The first thing you learned leading the People function at a product company was don’t break the machine. The machine was the product development loop: engineers, designers, and PMs converting salaries and compute into features, continuously. Everything else in the organization existed in relation to that loop. HR’s job was to feed it with talent, retain it with culture, and stay well clear of the process itself. Production was expensive, slow, and fragile. Breaking it cost more than anything HR could deliver by touching it.

That constraint shaped how people leaders thought about their role. Strategy meant staying aligned with what product needed. Impact meant keeping it moving. The best HR leaders in high-growth companies weren’t the ones who redesigned the machine. They were the ones who understood it well enough to get its requests fulfilled and mostly get the hell out of the way.

AI has collapsed that logic. Production is now cheap, fast, and increasingly automated across engineering, marketing, GTM, and operations. The constraint has moved into the messy domains people teams have always struggled to influence: how decisions get made, how change gets absorbed, how to revise the governing assumptions that shape how all teams operate. The imperative has inverted. We can no longer leave the machine alone. We have to break it and rebuild.

This piece is about what that rebuilding will take.

I. Execution Is Collapsing
What AI is compressing across functions as production ceases to be the primary constraint.

II. Constraint Migrates Upstream
Why faster individual output doesn’t produce faster organizational outcomes.

III. The Constraint Roadmap
How coordination, observability, and adoption absorb AI gains before they reach organizational output.

IV. When the Frame Becomes the Constraint
How governing assumptions become a ceiling AI cannot raise.

V. The Politics of Frame Revision
Why revising the frame requires absorbing real loss and political cost.

VI. How the Frame Changes
The operational sequence that makes revision politically survivable.

VII. The New Speed
What determines advantage at each stage of constraint collapse.

VIII. What Comes Next
How agentic systems may compress organizational structure, and why revision capacity has to be built now.

I. Execution Is Collapsing

A November 2024 Lenny's survey of 1,750 product managers, designers, and engineers showed AI primarily used to produce artifacts. Half of engineers were using it to write code, a third of founders for decision support, over a fifth of PMs and designers for documentation and research. This was the most AI-forward cohort available. Even here, respondents were saving around four hours per week.

One month later, Andrej Karpathy marked a new threshold:

“Coding agents basically didn’t work before December and basically work since — the models have significantly higher quality, long-term coherence and tenacity and they can power through large and long tasks, well past enough that it is extremely disruptive to the default programming workflow… You’re not typing computer code into an editor like the way things were since computers were invented, that era is over. You’re spinning up AI agents, giving them tasks in English and managing and reviewing their work in parallel.”

Peter Steinberger, creator of OpenClaw, described on The Pragmatic Engineer what this looks like operationally. Agents compile, lint, test, and validate their own work without constant human intervention. Prompts and plans replace pull requests as the durable unit of work. Multiple agents run simultaneously while humans scope tasks, sequence work, and resolve trade-offs. Code production ceases to be the primary constraint in engineering.

Outside engineering the same compression is underway, though unevenly. In execution-heavy, measurement-rich environments the production constraint has disappeared. Outbound prospecting, list enrichment, sequencing, and paid media optimization are now tractable with agents. The prospecting and sequencing layers of BDR work can be scaled at near-zero marginal cost. Performance marketing already operated inside algorithmic systems, and AI closes the remaining loops on creative generation, variant testing, and conversion optimization. Campaign production, audience segmentation, and performance reporting have compressed across marketing.

Enterprise sales has been protected by the layer that agents cannot replicate: relationship capital, credibility, and judgment accumulated over years. But the administrative layer surrounding those cycles is being stripped away. Walmart uses Pactum to run autonomous supplier negotiations across thousands of contracts simultaneously. As buyer incentives steer toward automation, sellers will follow.

In HR, the compression is similar but slightly delayed. AI sourcing tools run continuous candidate identification, enrichment, and outreach sequencing without recruiter throughput. AI screening now handles first-round qualification at scale, conducting, scoring, and scheduling conversations without human involvement until later stages. Onboarding platforms automatically resolve new hire document completion, system access requests, and day-one logistics. Frontline support tools now resolve the vast majority of employee and manager service requests without escalation. Policy and offer letter drafting runs from prompts if not triggered from system stages automatically. The execution layer of the people function is collapsing. The people function’s identity and operating model were built around supporting this execution.

For most people leaders, even in companies that describe themselves as AI-native, the world still looks surprisingly similar to a year ago: a few GenAI tools, a lot of pressure to produce an AI strategy, and a function that still runs on the same throughput model it always has. That window is closing.

II. Constraint Migrates Upstream

By mid-2025, AI was writing around 80% of code at Anthropic, a threshold most engineering organizations are still approaching. This did not make software production frictionless. It increased throughput significantly and at the same time showed where the next constraints would bind further growth.

Mike Krieger, Anthropic’s CPO, described what followed:

“We very rapidly became bottlenecked on things like our merge queue — the line to get your change accepted and deployed. We had to re-architect it because so much more code was being written. I started finding new bottlenecks everywhere. There’s an upstream bottleneck in decision-making and alignment. Then, once building is happening, there are coordination bottlenecks — making sure teams don’t step on each other’s toes. And when work is ready to ship, there’s air-traffic control: landing changes, coordinating releases, figuring out strategy.”

Krieger explicitly describes Goldratt’s central insight from The Goal: system throughput is determined by its slowest constraint, not by the average performance of its parts. Further pressure increases how much the new constraint binds. Anthropic didn’t solve organizational speed by automating code production. It revealed the next set of constraints by removing the one that had been masking them.

Anthropic is the most favorable possible case for this transition: hypergrowth trajectory, highest AI adoption in the industry, ideologically aligned leadership that understands the problem as it unfolds, no legacy systems accumulating drag. If constraint migration is this visible and this disruptive there, it will be slower to recognize and harder to resolve in organizations carrying heavier architectures and denser interdependencies. What Krieger describes is a structural feature of scale.

The Hierarchy of Constraints

Krieger’s bottlenecks are not random. They follow a predictable sequence that applies across every scaled organization attempting to absorb execution abundance.

All five constraints are present, though only one is binding at any time. As execution accelerates, pressure moves upstream. Resolving the binding constraint releases more outcome gains until the next constraint in the hierarchy binds.

Subscribe now

Why Scale Changes Everything

At small scale, the five constraints are tightly coupled to the workflow and team running it. A founding team of ten has no merge queue problem, no coordination overhead worth naming, clear and shared signal on what is working, no adoption lag between decision and behavior, and an governing assumptions that can be revised in an afternoon. Individual execution gains compound directly into organizational speed because the other four constraints are resolved quickly around the normal workflow.

Scale breaks this in a specific way. As organizations grow, three things happen simultaneously. Legible systems fragment: early signs that something is off no longer get picked up through shared context before they have to be named and escalated, and by then the insight has to travel through reporting layers that filter much of it out. Insight and authority drift apart: the people closest to the evidence lose proximity to the people with the power to act on it, and surfacing contradictions carries increasing cost as hierarchy steepens. Momentum accumulates: past decisions embed into architecture, hiring profiles, and capital allocation, making reversal a structural problem rather than a conversational one. The constraints were always there. What scale does is sever the connection between where they bind and where the capacity to resolve them sits, turning latent friction into active bottlenecks.

Lenny’s survey split between small and large companies suggests this decoupling is already underway. In smaller firms, execution gains still compound through the system. In larger ones, coordination, review, and approval overhead absorb them before they register as organizational output. The constraint has migrated.

III. The Constraint Roadmap

The next two to five years of organizational transformation will be shaped less by emerging model capabilities than by constraint migration. As execution collapses across functions, most firms are layering AI onto 1x infrastructure. Coordination, observability, and adoption become the binding constraints. AI multiplies artifact volume, but routing structures, attribution systems, and absorption capacity have not kept pace. Working through these constraints in sequence becomes the de facto organizational transformation roadmap.

Coordination

Coordination is the friction generated by interdependencies across teams as work is planned, built, and delivered. It scales with the number of interdependencies, not the volume of work. That distinction is what makes it hard to diagnose correctly.

When execution accelerates and organizational speed doesn’t follow, the instinctive response is more alignment: more meetings, more documentation, more cross-functional syncs. These address the symptom without touching the structure. When production speeds up, the volume of work requiring coordination grows faster than output does. As Rizwan Iqbal, a former colleague and now Director of Engineering at Superhuman, put it:

“Most of the time is spent orienting: understanding the system, aligning with other teams, and figuring out where changes are supposed to live. That’s not a tooling problem. It’s a routing problem.”

For organizations with clean ownership and explicit protocols, AI improves capture, routing, and orchestration. For those without it, AI accelerates documentation of the same underlying confusion. The tool moves faster, but the routing logic remains unchanged.

As agentic systems scale, the problem compounds. Cory Ondrejka, former tech advisor to Sundar Pichai, argues that the challenge of scaling agents mirrors that of scaling people as the number of interactions to be resolved grows faster than artifact production. Organizations become composite systems of humans and agents sharing resources, and coordination protocols that were previously socially inferred must become explicit.

Coordination does not require better alignment rituals. The structural fix is ownership clarity and modular architecture that reduce the number of interdependencies requiring active coordination. Where that structure exists, AI can help at three layers: converging the additional signals it generated as work scales, facilitating human decision-making both procedurally and substantively, and resolving some dependencies directly through agent-to-agent coordination.

Observability

Observability is the system’s ability to see the consequences of its own actions clearly enough to understand what changed and why. That only matters if the organization can connect what it did to what happened next. Most organizations are full of dashboards and still largely unable to tell which initiatives caused which outcomes.

Charity Majors, CTO of Honeycomb.io, states the organizational implication directly:

“Your ability to get any returns on your investments into AI will be limited by how swiftly you can validate your changes and learn from them.”

The same principle that makes software systems adaptive applies to organizations as well: generate enough signal to understand what the system is doing, what it is not doing, and why.

When observability binds, the organization produces results it cannot learn from. The instinctive response is to add review gates: more approval steps, more governance checkpoints. These slow execution without improving clarity. The problem is not a lack of oversight. It is a lack of evidence linking actions to outcomes.

Observability varies sharply across functions and the gap matters. Product and engineering teams have continuous passive signals from production systems. GTM teams have partial but improving visibility through conversion data and pipeline metrics. People systems remain the least observable domain in most organizations: episodic surveys with multi-week lag, manager narratives that reflect what gets surfaced rather than the full reality, and platforms built for records management rather than diagnosis. Most organizations can observe their software and their customers far better than they can observe themselves.

As initiative volume rises, the interactions between them multiply faster than the work itself. Clarity degrades not because the signal disappears but because the ratio of signal to noise collapses under volume. The fix is to apply the measurement discipline developed in more observable functions to the rest of the organization: shaping how work is planned, scoped, tracked, and assessed so actions can be linked to outcomes. AI can help realize this through new forms of data capture, transcription, pattern detection, and cross-system synthesis that reveal how work is actually unfolding across conversations, workflows, and decisions.

Adoption

When execution becomes cheap, organizations become saturated with partially implemented work. Features are built faster. Campaigns are drafted and launched more quickly. New internal tools and workflows are spun up in days rather than months. Launch becomes easy, while engagement, activation, and implementation remain constrained.

Adoption capacity functions as a form of work-in-progress limit, borrowed from Kanban. Every organization has a finite amount of new behavior, new tooling, and new process it can integrate into real work at once. Execution abundance increases pressure from every direction: leaders push transformation programs, cross-functional teams introduce new workflows, central functions deploy company-wide initiatives, and vendors release new capabilities into tools the organization already uses.

As the limit is consistently exceeded, partial implementation becomes the norm. Employees lose clarity on the organization’s actual priorities. Leaders struggle to reconcile competing initiatives or determine which work matters. Operational overload produces a predictable sequence: overwhelm, then cynicism, then compliance without integration. Each failed round increases the skepticism applied to the next.

Adoption capacity can be raised. AI coaching reduces the cognitive overhead of applying unfamiliar tools and processes under real conditions. LLM-connected knowledge systems help employees build a clearer mental model of the firm, its active initiatives, and how their work connects to them. The draw on capacity can also be reduced. Roadmapping against adoption capacity rather than build capacity makes the WIP limit explicit and forces the prioritization conversation most organizations avoid. Modular initiative design reduces how much capacity each change consumes by breaking programs into independently adoptable components.

These three constraints define the coming agenda for the people function: enabling coordination at 5x volume, building the observability infrastructure required to learn at production scale, and raising the adoption ceiling so throughput gains actually compound. This is the work inside the engine.

IV. When the Frame Becomes the Constraint

Every organization runs inside an operating frame: the core assumptions defining what it optimizes for, how decisions get made, what gets resourced, which structures are treated as fixed, and what behaviors get rewarded. Execution and the constraints that follow all operate inside this organizational solution space. They cannot change it. As the lower constraints are resolved, optimization within the frame continues until the frame itself becomes the constraint.

AI collapses the cost of optimizing within this frame. Execution accelerates, and every capability gained from the prior three constraints compounds inside the same operating frame. Chris Argyris called this single-loop learning: detecting errors and making improvements in workflow within an existing set of governing variables. He differentiated it from double-loop learning: surfacing, challenging, and replacing the governing variables themselves. AI removes friction from single-loop optimization. It does not initiate double-loop learning. Teams move faster inside the frame while the frame itself remains unchanged.

Competence accumulates around the frame’s assumptions. Success inside a logic reinforces confidence in the logic that produced it. Hiring profiles, incentive structures, and institutional memory encode those assumptions as defaults. The ceiling does not announce itself. It presents as a strategy question, a resource question, a talent question. By the time it is legible as a frame problem, the cost of revision has compounded.

Frame revision is not a calibration. It requires new products, market repositioning, architectural restructuring, operating model redesign, and terminating legacy work. These are not optimization moves. They require revising what the organization is for, how authority flows through it, and what prior bets it is willing to name as wrong. That is a different order of problem from anything the first four constraints demanded.

Meta’s AI pivot shows what changing an established frame can take. Through 2022 and 2023, Meta had publicly staked its AI positioning on open-source development and an academically oriented research culture, with Yann LeCun’s public skepticism about the LLM path to AGI defining its scientific identity. By 2024, closed frontier model development had pulled decisively ahead on capability. Llama 4’s April 2025 launch underwhelmed. By June, Zuckerberg had publicly expressed frustration.

Within a single quarter, Meta replaced governing assumptions of open-source research, academic ethos, and distributed decision-making with concentrated authority, closed development, and commercial deployment speed. It dissolved its existing research lab structure, consolidated operations under Superintelligence Labs, offered compensation packages reaching $100M to recruit researchers from OpenAI, Google, and Anthropic, and acquihired Alexandr Wang from Scale AI in a $14.3 billion deal. In November 2025, LeCun left Meta entirely. A company of Meta’s scale revised the governing logic of its AI division within a single quarter of committing to the shift. Most organizations of this size take that long just to name the problem.

As AI capabilities become broadly accessible, execution advantages built on them decay faster. Features that once required months of engineering effort can now be rebuilt by competitors in weeks. The cost of replication falls alongside the cost of production, so improvements diffuse quickly across competitors operating inside similar frames. Firms converge toward similar outputs not because they choose the same strategy, but because similar optimization logic produces similar ceilings. Optimization inside a saturating frame produces the unusual outcome of products improving even as differentiation narrows.

V. The Politics of Frame Revision

Revising the operating frame redistributes budget, scope, status, and future upside across identifiable people before it produces anything new. The leader whose team gets defunded, the executive whose prior bet gets named as wrong, the function whose authority shrinks as control shifts: these are not casualties of a strategy process. They are the people most invested in the assumptions being revised, and they sit inside every conversation where revision has to happen.

Working within the frame, political cost is manageable. Coordination redesign may create winners and losers, but the losses are bounded and the benefits are clear. Frame revision is different: the losses are structural, and the people being asked to absorb them must accept certain costs against uncertain returns. That asymmetry is what makes it categorically different from every constraint below it.

Each phase of revision carries a distinct political cost of participation. Surfacing frame misalignment before certainty exists requires tolerance for vulnerability. Admitting that existing logic may no longer fit signals loss of control before any alternative is ready. Naming the assumptions tied to identity, authority, and prior bets accepts questioning. In organizations that reward decisiveness this reads as weakness rather than rigor. Reallocating capital and authority to enforce revised priorities requires taking visible loss, producing identifiable winners and losers with named sponsors and making the political cost concrete and attributable. Embedding new defaults under real operating pressure requires allowing for experimentation, because acting differently under real stakes risks short-term variance in systems against a culture where stability is what gets rewarded. These are the structural costs of revision.

These political costs activate defensive routines across the organization. Chris Argyris observed that under threat, intelligent professionals consistently protect prior reasoning from public examination. Goals are defined unilaterally, conversation is controlled to avoid exposure of error, and structural tension is reframed as execution failure rather than examined at the level of assumptions. It is not incompetence or bad faith but a learned survival strategy in performance-driven systems. Under threat, professionals protect the reasoning that earned them authority, applying the same analytical capability that made them credible in the first place.

These routines are self-sealing. When a strategic contradiction surfaces, it is delegated to execution: resourcing, sequencing, communication, talent. The governing assumptions remain untouched. Single-loop errors are visible, discussable, and correctable. At the level of governing variables, defensive narratives justify themselves by explaining away disconfirming evidence. Governing assumptions become non-negotiable, non-testable, and often invisible, raising the political cost of revision.

Whether a firm can successfully revise its frame depends on its capacity to absorb the political costs of revision. That capacity rests on authority to change priorities and rebalance power, legitimacy to admit that prior assumptions no longer fit, and the ability to enforce revised assumptions through real trade-offs in capital, headcount, and governance. When that capacity falls short, declared change stalls and the prior frame reasserts itself through the normal incentives and routines of the organization.

AI can make strategic contradiction much easier to see. It models opportunity cost, surfaces inconsistencies between stated strategy and actual allocation, and makes misalignment legible. For leaders with authority to act, this reduces the time between detection and articulation. But AI cannot manufacture legitimacy, absorb structural loss, or enforce revised assumptions. It strengthens leaders who can already act and leaves unchanged the political conditions that determine whether they will.

Founder-mode environments are often described as low-politics because they demand strong alignment around governing assumptions. But politics does not disappear. It concentrates vertically around identity and proximity to the founder. The assumptions most resistant to revision are often the founder’s convictions, especially the ones that helped scale the company. In practice, the people with the clearest view of disconfirming evidence are often furthest from permission to surface it, because the political cost is correctly perceived as high.

VI. How the Frame Changes

The operating frame is composed of both explicit and tacit knowledge. The explicit forms are visible: stated strategy, resource allocation, org structure, and the metrics that define what counts as success. The tacit forms are not: the judgments that feel like common sense, the practices that encode prior decisions long after anyone remembers choosing them, and the interpretive habits that determine which signals are taken seriously and which are explained away. Without tacit change, we have slideware. Without explicit change, revised assumptions never reach frontline decisions. Revising the frame requires converting between these forms across individuals and systems: surfacing tacit assumptions into something explicit enough to examine and contest, then embedding revised assumptions back into the judgments and practices that shape how work actually happens.

Argyris helped us see why revising the operating frame generates political cost. Nonaka and Takeuchi show how treating frame revision as an organizational learning process can reduce that cost. Their SECI model (Socialization, Externalization, Combination, Internalization) describes the sequence by which experience becomes structural change. Political friction impedes each step of knowledge conversion. Accelerating the flow reduces its effective force, not by eliminating cost but by moving shared understanding faster than defensive routines can organize.

Socialization

Socialization converts private experience into shared situational awareness. When people exposed to similar consequences begin to sense that existing assumptions no longer fit, that recognition spreads through proximity and repeated exposure before it is formally articulated. This broader awareness lowers the political cost for those who later surface misfit before certainty exists. Naming a problem without a solution, before any alternative is ready, is the moment of maximum vulnerability. When recognition is shared rather than private, revision lands on experience people already hold instead of as an abstract challenge to authority. Socialization does not decide anything. It makes articulation survivable.

Socialization is strengthened by widening shared exposure to real work and real consequences: collaborative workflows, recorded sessions, customer calls, behavioral data, and feedback signals that spread tacit understanding without requiring direct participation in every interaction. AI reduces the cost of aggregating and replaying this material at scale, widening access to consequences across the organization. But shared access is not the same as shared experience. Tacit recognition spreads through repeated, socially situated exposure to consequences, not through information distribution alone.

Externalization

Externalization converts shared awareness into explicit, contestable assumptions. What was previously sensed becomes something that can be named, examined, and revised. An organization cannot change what it cannot name. Governing assumptions carry taboo because making them explicit exposes what prior decisions were built on and whose authority those decisions reinforced. Structured discussion distributes that exposure across the conversation instead of concentrating it on the person who speaks first. That is the political cost Externalization reduces: not the cost of being wrong, but the cost of being the one who says it.

This phase is strengthened by institutionalizing forums where competing interpretations can be contrasted, assumptions surfaced, and trade-offs made explicit without immediate penalty. AI can support this phase by surfacing inconsistencies between stated strategy and actual allocation, prompting assumption articulation, and synthesizing divergent interpretations without deference to status or taboo. It cannot remove the political cost of the response once assumptions tied to identity, authority, or prior bets are made explicit.

Combination

Combination converts competing interpretations of revised assumptions into aligned strategic direction. In this phase, frame revision becomes consequential through explicit trade-offs across roadmaps, resource allocation, and governance. The shared awareness and conceptual clarity of earlier phases provide the rationale required for these difficult choices. Programs are terminated. Resources move. Decision rights shift. When trade-offs are explicit and integrated rather than implicit and fragmented, loss is legible. This is more absorbable than loss that feels arbitrary.

Direction stalls when plans are revised but budgets, incentives, and governance remain untouched long enough for the existing allocation of capital and authority to continue governing by default. AI compresses this phase by rapidly generating and stress-testing coherent, cross-domain plans, modeling opportunity costs, and surfacing trade-offs that would otherwise remain politically unsayable. It lowers the friction of producing integrated direction. It cannot absorb the political cost of the trade-offs that direction requires.

Internalization

Internalization converts articulated direction into organizational capability. Explicit principles, decision rules, and strategic shifts become patterns of action under real operating pressure. It completes the frame revision when local actors make trade-offs consistent with the revised frame without consulting the document, because the logic has become reflex. This phase reduces the political cost of enforcing the new frame. Distributed experimentation seeds examples of the revised assumptions applied under real conditions, building the social proof that pulls adoption rather than requiring it to be mandated. The more internalization spreads through example, the less authority has to be spent enforcing compliance.

AI accelerates Internalization at the level of individual practice. AI coaching embedded in decision-making and reflection reduces the cognitive overhead of applying an unfamiliar frame under pressure. Surfacing examples of the revised logic applied well creates social proof across distributed teams without central curation. Behavioral drift detection gives leaders earlier signal when local decisions revert to prior frame logic, reducing enforcement cost before that drift hardens into a norm.

The end of one cycle seeds the next. Internalization completes as the revised frame becomes reflex, and that reflex immediately generates new tacit experience. Revised assumptions surface new constraints, create new misalignments, and expose trade-offs that were previously invisible. That experience becomes raw material that Socialization converts in the next round. This frame revision cycle once took years in scaled firms. Execution-abundant markets are compressing it to months. The organizations that thrive are not those that get revision right once, but those that build the capacity to keep revising.

Enabling Conditions for Revision

The revision cycle runs on accurate perception. For signals to challenge the frame rather than be absorbed by it, two conditions have to be in place. Without them, Socialization spreads noise, Externalization devolves into opinion, Combination resolves around power, and Internalization reinforces the prior frame.

The first is falsifiable signal: evidence designed to detect when governing assumptions are wrong, not merely when execution is underperforming. This is a higher bar than the operational observability described earlier, which indicates whether initiatives are producing results. A falsifiable signal indicates whether the assumptions underlying those initiatives remain valid. An organization can have excellent dashboards and zero falsifiable signal if everything it measures is optimized inside the frame. Nothing it tracks can indicate that the frame itself is misaligned. The strongest incentives are to avoid generating the one piece of evidence that could falsify the current strategy. The survey that would test the core assumption doesn’t get sent. The experiment that could disprove the thesis doesn’t get funded.

The second is interpretive talent: people capable of reading misfit signals at the level of governing assumptions, rather than translating it immediately into an execution fix. That requires domain fluency to distinguish meaningful deviation from noise, systems thinking to reason about feedback loops and second-order effects, an improvement orientation that treats outcomes as provisional rather than as verdicts, and metacognition to notice one’s own framing under pressure before defending it. Select for people who revise their own governing assumptions in the same way the organization must revise its own.

VII. The New Speed

Execution speed is the most visible benefit of AI. It produces measurable output, aligns with how individuals experience the technology, and generates immediate evidence of progress. That visibility makes it easy to measure and easy to value. It does not make it decisive.

Every organization in this audience is already moving faster than it was two years ago. Execution is accelerating across engineering, GTM, marketing, and operations. The gains are real, but they are increasingly mistaken for durable advantage.

As AI capabilities spread across competitors, execution speed no longer sustains differentiation. The cost of replicating what competitors build is falling alongside the cost of building it. Advantages that once lasted multiple planning cycles now compress toward parity within one. Throughput is now the entry condition for competing, not the variable that determines competitive distance.

Each organization needs to identify which speed addresses its binding constraint. Many companies are still barely using AI to improve how fast work gets done. For them, execution speed is binding, and improving it determines whether the organization stays default alive in a market where baseline progress is no longer optional. For those that have optimized execution but are not seeing outcomes follow, constraint resolution speed is binding: how fast coordination, observability, and adoption convert execution gains into sustained quality leadership until competitors similarly optimize. For those where outcomes are compounding but differentiation is narrowing, frame revision speed is binding: how fast governing assumptions change in response to disconfirming evidence, and whether those outcomes sustain differentiation or compress toward commoditization.

High throughput feels like organizational health. Smooth constraint resolution feels like winning. Both can be true while the frame that generates them races toward commoditization. The metric that signals success at one speed obscures the binding constraint at the next. It's not just a question of how fast, but of which speed matters. What follows is a diagnostic for each speed: internal signals that reveal whether a constraint is resolving, and external signals that reveal whether resolving it is still where competitive distance is being created.

Gear 1: Execution Speed

When an organization is producing below the baseline the market now requires, execution speed is the highest-leverage focus. Individual contributors are using AI to increase task throughput without that gain transforming team-level output. Work is produced faster and artifacts multiply, yet delivery lead times, backlog depth, and team-level output capacity remain unchanged.

Two measures track execution speed improvement. Cycle Time measures the interval from intent specified to artifact produced at quality standard: code shipped, campaign launched, candidate assessed, contract drafted. Process Completion Time measures the recurring processes the organization depends on: hiring loops, onboarding, campaign production, deal close. Both should compress as AI deployment reaches the team level. If they have not, individual productivity gains have not translated into team throughput. Execution remains the priority.

The external signal is in contrast to market delivery expectations. Is the organization shipping the features, practices, and pricing models that have become the market baseline? If not, execution speed remains the priority. If so, the constraint may have migrated.

Gear 2: Constraint Resolution Speed

When AI has transformed team throughput but organizational outcomes are not following, constraint resolution speed unlocks the next wave of competitive gains. The coordination, observability, and adoption constraints covered earlier in this piece are what bind at this stage. Execution gains are being absorbed before they reach organizational output.

Absorption Latency measures the time from when a change is directed to when new behavior appears consistently in operational data. This might be the interval between a new architecture decision and seeing it reflected in code structure, between a revised ICP and seeing it reflected in which deals get pursued, or between a new performance standard and seeing it reflected in promotion decisions. Tracking adoption without a quality standard mistakes coordination failure for progress. Most organizations cannot measure the start of absorption directly, because the moment a change is directed is not well recorded. In practice, they use deployment date as a proxy. The difference is that absorption latency begins when the organization commits to the change, while the proxy begins only once the change is launched. If the measured interval is not shrinking as execution volume rises, constraint resolution speed remains the priority.

The external signal is replication speed. Can the organization copy what competitors are building as fast as they can copy you? If competitors replicate your capabilities faster than you can match theirs, or you are consistently behind on features and practices market leaders have already validated, constraint resolution speed remains the priority. If replication is roughly symmetric, or you remain the sustained quality leader in the market, the constraint may have migrated to the frame itself.

Gear 3: Frame Revision Speed

When the organization’s optimized operations are producing market leadership but differentiation is narrowing, frame revision speed is what sustains competitive advantage. The governing assumptions directing that execution are producing a ceiling competitors across the market will reach at roughly the same time. The frame is the constraint, not the performance.

Revision Latency is the primary measure: the interval between the first observable structural contradiction and a measurable shift in capital, headcount, or architecture. The first timestamp is when disconfirming evidence becomes observable. The second is when allocation actually moved. AI analysis of collaboration data, planning artifacts, and allocation records can surface both. Without it, the first timestamp is reconstructed in hindsight, and the lag shrinks on paper rather than in reality.

Strategic Kill Rate and Reallocation Magnitude are the enforcement signals that confirm whether revision is structural or just declared. Kill Rate measures how frequently coherent work gets terminated because it no longer fits the governing frame. A low rate may indicate the organization cannot absorb the political cost of stopping work, leaving capital distributed across initiatives that no longer serve the governing logic while new priorities go underfunded. Reallocation Magnitude confirms the shift: the percentage of budget and talent density moved toward new governing variables within a planning cycle. Without it, revision latency is symbolic and the prior logic governs by default regardless of what the strategy document says. Who took the loss this quarter that proved the new direction was real? Most leadership teams cannot produce a name. When they cannot, the frame remains intact.

The external signal is a market trending toward commoditization. Competitors are trending toward outcomes of similar quality at similar cost, pricing power is eroding despite strong delivery, and category boundaries are shifting as competitors reposition or redefine what the market competes on. Successful frame revision opens new terrain to optimize against. But the firms that sustain advantage are not the ones that revise once. They revise faster than the market can converge on what they’ve built.

VIII. What Comes Next

This piece is written for the current interval, between AI-enabled work collapsing execution and agentic systems replacing the organizational structure itself. How long that interval remains open is uncertain. It may compress into two years or stretch to five.

HPP8: ‘The End of the Organization’ will speculate on what transitioning to agent-centered organizational structures means for competition, employees, and the people function. Frontier models arriving in the next 18 months would make agentic systems robust enough to deploy across most knowledge work functions. At their production volumes, the constraints this piece describes would bind simultaneously rather than sequentially. Agentic systems would also begin absorbing friction in coordination, observability, and adoption, compressing the hierarchy into months. As execution and constraint-resolution systems integrate, they can produce an operational flywheel with a speed and responsiveness that make the old organizational structure nonviable. That transition would fire a starting gun for a new stage of competition.

What this piece establishes is the prerequisite for this transition. The firms that work through the constraint hierarchy sequentially arrive at that starting gun with revision machinery already operational. The firms that don’t will face the stacked version: frame revision at maximum scale, on maximum assumptions, under competitive pressure, without the capacity the sequential path would have built. The race is lost before the starting gun fires.

For SaaS founders

You are already in frame revision, but you probably haven’t named it that way. The coordination and adoption work you are doing is important, but it operates inside a frame that the market is making obsolete. The market is telling you seat-based pricing no longer reflects the value you deliver, but repricing risks the revenue engine. The market wants AI automation but your management structures were built for human decision chains, and the people who built those structures are still in the room. Your product roadmap includes agentic features but your enterprise security posture was designed to be differentiating. None of these are product problems or go-to-market problems. They are changes to the governing assumptions that define your solution space, running without the revision infrastructure this piece describes.

The question is how you build the capacity to take on larger frame revisions before the market forces them on you.

For AI-native founders

The frame you are building now is encoding itself into every hire, partnership, and architectural decision. Execution is your advantage and it has been compounding. You are likely working through coordination and observability limits but the frame those constraints operate within is hardening faster than you can see it. The concentrated authority and conviction that makes disruption possible makes self-revision hard. You built for differentiation, assuming you would remain the differentiator. Your pricing model, talent structure, and market position all express that assumption. How long before a new team, less burdened by your accumulated technical and organizational debt, takes the next model and builds 70% of what you have created?

Founder control can absorb one revision. It does not create the distributed capacity to sustain repeated revision under pressure. The infrastructure required is organizational and political, not positional. The question is whether you are building it now, or whether each successive revision will cost more than the last because the organization was never designed to bear it.

Architect for the Organization

In 2014, I joined RelateIQ for my first Head of People role. Steve Loughlin, then CEO, told me I needed to be Product Manager for the Organization. I nodded in knowing agreement, understanding nothing. It took six years to work through what that actually meant: building a roadmap executives actually owned, scoping problems before designing solutions, and launching with leaders rather than at them. Getting the machine to run well. That was the right job for the constraint it addressed. The constraint has moved.

Previously, we assumed the operating frame was correct and needed to be operationalized and protected. Now we must assume the frame will require revision. The work becomes architectural: designing the conditions under which governing assumptions can change while execution continues to accelerate across every function the people team supports. The classical mandate to Hire, Motivate, and Retain is superseded by a new one: Accelerate, Release, and Revise.

The agentic transition concentrates its hardest problems in organizational development: not adjacent to OD mastery but identical to it. The people executive must guide this transition not because the role confers authority, but because the work requires expertise in navigating the political dynamics that determine whether revision is survivable, building the organizational learning processes that make it possible, and driving the behavior change that makes it real. That expertise has to be built, not assumed, and the learning curve gives months, not years.

Most organizations get stuck in activity overload. They run engagement surveys, performance reviews, reorgs, and training initiatives, yet the business does not meaningfully improve. In top scale-ups, the issue is not effort. It’s that much of this work isn’t connected to how the company actually creates value.

This keeps leaders from chasing scattered HR programs and helps them invest in a small number of targeted, testable changes tied directly to performance.

How the Process Works in Four Steps

1. Define the Business North Star

Core Question: What is the single business outcome that matters most in the next 12–24 months?

A North Star is the measure of the value the business must deliver. It guides prioritization and trade-off decisions for the leadership team. It’s too high-level and lagging to tie directly to People Initiatives, but it sets the essential scope for the rest of the chain.

Common examples:

• Net new ARR

• Customer retention

• Product Usage

Output:

• A short North Star statement with one or two clear metrics.

2. Map the Value Streams

Core Question: What are the few core flows of work that contribute most directly to the North Star?

Value Streams are the cross-functional flows of work that show how the business actually creates value and what that work produces. They make clear the outputs that must be delivered, at the right quality, frequency, and scale, to achieve the North Star.

People and org changes only create impact when they improve these flows. If you don’t understand the work, you can’t design changes that improve it.

Common examples:

1- Product Development:

• This stream evaluates how effectively the company turns opportunities into shipped, adopted product that delivers value.

• Process Stages:

  • Discover & Prioritize: Product identifies customer problems, sizes opportunities, and ranks work by impact and confidence.

  • Design & Specification: PM and Design create flows, prototypes, requirements, and acceptance criteria.

  • Build & Test: Engineering develops, integrates, and tests the feature; QA ensures quality and stability.

  • Release & Monitor: DevOps and Release Engineering deploy; Product monitors quality, adoption, and performance.

  • Launch & Enable: Product Marketing communicates value and equips sales and support.

• Workflow Outputs:

  • Feature meets acceptance criteria and quality bar

  • Solution validated against stated user need

  • Clear insights produced for the next iteration

• This flow reveals where prioritization breaks, specs lack clarity, engineering is blocked, or release processes create friction.

2- Acquire & Close:

• This stream evaluates how predictably and efficiently the company converts market demand into high-quality revenue.

• Process Stages:

  • Demand Generation: Marketing generates inbound and outbound interest.

  • Qualification: SDR/BDR validates fit, timing, authority, budget, and need.

  • Solutioning & Proposal: AE shapes requirements, builds the value case, models pricing and margins, and validates feasibility with delivery teams.

  • Closing & Contracting: AE, Legal, and Finance finalize terms and economics.

  • Handoff to Delivery / CS: Expectations and success criteria transfer cleanly.

• Workflow Outputs:

  • Data-rich pipeline with accurate forecasting

  • Precise scoping and pricing with clear margin expectations

  • Clean handoff with aligned expectations and no operational surprises

• This flow highlights where pipeline quality drops, deals get mis-scoped, or cross-team handoffs create churn.

Output:

• A simple map of 2–3 value streams with defined outputs and current breakdown points. Multi-product or complex organizations may have more, but focus on the fewest necessary.

• Most of this information already exists inside the company. The process surfaces it through structured discovery.

3. Diagnose the People Levers (What’s Blocking the Work)

Core Question: Which People Levers could most improve how the Value Streams produce their outputs?

This step surfaces the people constraints that slow or distort the Value Streams. It shifts the conversation from high-level solutions (“we need manager training” or “we need to clarify ways of working”) to real workflow issues (“handoffs break because ownership is unclear” or “managers cannot coach because their spans are too wide”).

Each constraint sits in one of the four areas below, covering two dimensions (Systemic–Behavioral and Group–Individual), which helps us see where constraints originate.

Subscribe now

For each Value Stream, we walk through all four areas, map the influencing Levers, and then prioritize the few that would make the biggest difference. This focus becomes our People Strategy.

A. Structure

How the organization is set up to support the work.

1. Team and pod design

2. Reporting lines and spans of control

3. Role clarity, accountabilities, and decision-rights

4. Cross-functional interfaces

Structural issues often show up as slow decisions, unclear ownership, duplicated work, or leaders stretched too thin to coach.

B. Talent Density

Who is in the seats, and what they can actually do.

1. Leadership and manager capability

2. Skill gaps that need to be filled

3. Hiring needs

4. Managing out to raise the quality bar

Talent constraints appear as repeated errors, weak judgment calls, or leaders unable to run the pace the work requires.

C. Collaboration

How teams coordinate and make decisions together.

1. Alignment and review cadences

2. Handoff practices and issue-surfacing

3. Communication and knowledge-sharing

4. Goal clarity and connection to outcomes

Collaboration constraints usually show up as friction between teams, slow escalations, mismatched expectations, or teams working at cross-purposes.

D. Culture

The motivation, ownership, and energy needed to sustain the work.

1. Connection to purpose

2. Drive for the work / Burnout

3. Trust across teams and in leadership

4. Accountability and outcomes

Cultural constraints often appear as low follow-through, weak standards, inconsistent accountability, or teams doing the minimum needed.

Output:

For each Value Stream:

• Create a People Strategy statement for each Value Stream: “We will focus on [A, B, and C specific aspects of the levers], to achieve [desired change in Value Stream output] in [timeframe]”.

4. Build the Executive Roadmap of People Initiatives

Core Question: What high-impact work must the Leadership Team align on to shift the Levers and improve the Value Streams?

Our People Strategy informs the People Roadmap: the set of initiatives needed to move each of the Levers we prioritized. Each initiative is scoped to improve a specific part of a Value Stream. That level of clarity helps us target, assess, and iterate until the workflow actually improves.

The Roadmap should be shared across the Leadership Team, reviewed in executive meetings, and updated monthly. It shows leaders where their involvement is needed, what trade-offs they’re making, and what will or won’t ship. It also creates order in the parallel work that growing companies rely on, making People and Org Design efforts feel more coordinated and predictable.

For a deeper look at applying product thinking to People work, here’s more detail from me: hardpeopleproblems.com/p/people-product-managers-playbook

Output:

• A People Roadmap for the current quarter and the two that follow, in a format reviewable by the Leadership Team in a 15-minute agenda item.

• Each Initiative on the Roadmap is scoped well to:

  • show how it will improve a Value Stream,

  • identify its main users or internal customers,

  • the “lift” level needed from the Leadership Team, and

  • indicate its current stage (Discovery, Design, Build, Launch).

• The Leadership Team is aligned on the Roadmap’s priorities, the impact each item aims to achieve, and the role they play in delivering it.

What You Leave the Process With

1. A Shared View of How People and Org Design Shapes How the Business Really Works

The process gives the company a clear, shared map of how People factors shape the work across Product Development, Go-to-Market, and Operations.

• For the People team, it builds a deeper grasp of the business mechanics so solutions are grounded in real work, not generic best practices.

• For the Leadership team, it reveals how People and Org Design choices directly affect speed, quality, and scale.

2. The clarity to prioritize real business impact — and the ability to keep work anchored to it

The method aligns the company on three layers of impact: the North Star, the outputs of the Value Streams, and the critical workflows within them. This makes it easier to see which changes matter most and where to focus time, talent, and investment.

Because every Initiative is scoped to improve a specific piece of the workflow, the work stays tied to business impact through design, assessment, and iteration. This prevents low-value activity and helps the company build faster by concentrating effort on the changes that actually shift performance.

3. A Leadership-Owned Plan for Driving the Changes That Matter

You finish with an Executive Roadmap that shows what will ship, when, and where leadership involvement is required. People initiatives only succeed when executives shape them early (so they’re relevant and realistic) and support them visibly at launch (so teams adopt them on Day One). This turns People and Org Design from an HR task list into a shared leadership agenda.

(5-minute podcast introduction, edited from Google’s NotebookLLM and Audiotease)

Around a third of major People team initiatives never ship. A third of what does ship is rejected by users on deployment. And a third of what engages users fails to move the needle for the business. This would mean that only three out of every ten People initiatives drive real business impact. It’s a harsh but realistic reality, given how poorly executives typically view HR’s effectiveness.

Executives are a unique combination of buyer, delivery channel, and user of People services. They can undermine an initiative’s impact simply by doing nothing. If they don’t own an initiative, their managers won’t. If managers don’t, employees won’t. And then, despite all the time invested in its development, the launch fails.

Yet, as People Leaders, we often try to shield executives from the “burden” of HR strategies and initiatives. The reality is it’s impossible to build and launch “organizational features”—onboarding, comp structures, leadership training—without deep executive alignment and ownership. We need a repeatable way to make it easy for executives to influence what we build and actively champion its launch.

Product teams have a solid toolkit for solving this problem. They have structured processes to engage stakeholders across the product planning and development cycle. Some of this product thinking, focusing on user experience, user data, and iterative design, has already reached HR. But we’ve barely scratched the surface.

This piece transfers two core tools from product management to scale-up HR—helping People leaders partner with executives to build and launch initiatives that actually land.

1. The People Roadmap – How to build and leverage a well-scoped organizational roadmap with your executive team.

2. The People Product Development Process – What touchpoints with executives are needed during scoping, design, and launch to ensure adoption and impact?

By considering the “organization like a product” with HR as the product manager, we can build new features that teams actually use—driving business growth, developing talent, and shaping an engaged workplace. Because at the end of the day, great organizations don’t just grow—they’re shipped.

The Roadmap

Geoff Charles, VP of Product at Ramp, shares how he uses roadmaps to align the leaders.

“We share a high-level roadmap, get feedback from leaders across the company, and ensure it is in lockstep with large marketing moments we co-develop. This helps us amplify our product releases and get loud in the market. Planning is not just specifying goals but also trade-offs. Everyone wants everything to be done—but that simply can’t happen. By making explicit what trade-offs you are making, we were able to engage in much more nuanced discussions around whether we should do X vs. Y instead of the absolute.”

We have similar needs for successfully launching our People products. Function leaders are our closest partners and often the most effective user marketing channel. Without their education and engagement, failure is inevitable. A shared roadmap ensures that executives know what products are being developed, why they are important, and where they can influence each feature. HR Business Partners can also use these plans to frame a shared agenda with their executive clients and position themselves as strategic advisors.

The Company’s People Roadmap should be a prioritized schedule of products needed to progress toward our goals and execute our People Strategy.

• Lenny Rachitsky, Lenny’s Newsletter, Oct 18, 2022

The most common mistake is having too much on the roadmap, which is evidence of an absence of the strategy needed to prioritize. Products on the roadmap should impact the business and require a significant lift from the executive team. In a previous article, I described how answering three questions can help form the core of a People Strategy closely tied to business impact.

If you don’t yet have a People strategy, the bulk of your roadmap should consist of products that answer these questions.

A People roadmap should align stakeholders on what products will/will not ship and their intended impact. It should also guide their involvement in each product's scoping, design, or launch stage. Like a customer-facing roadmap from the product world, a shared People Roadmap needs to be simple, with only the information required for executives to voice their feedback on impact and priority easily.

The Miro board I’ve created above simply shows the Product Name, Development Stage, Business Impact, and User Objective. The reverse shows details on the Launch Date, Responsibilities, and Lift Size for the Executive Team. The roadmap should be highlighted with each executive engagement and updated monthly. It should bring some calm to the chaos of having multiple products at different stages and the many roles executives are asked to play in our work.

Avoiding Product Failure Traps

A roadmap is critical, but the gap between intention and effect is massive. As mentioned, only three of every ten initiatives move the needle for the business. The others fail to engage users, make an impact, or even fail to ship. These three traps correlate with product success criteria from design thinking: feasibility, desirability, and viability. Understanding these anti-patterns can help inform a repeatable design and launch process to overcome them.

To show how these factors can be used in a design process, here is a simple framework I built to guide teams in designing small products that can be completed in a 2-week sprint. In this very light process, product plans are mapped against the Viability, Desirability, and Feasibility factors here, then assessed with their Team Lead to make a go/no go call before development starts. This could also be a good place to start for teams that find the following chapters overwhelming.

Subscribe now

Scope right until it hurts

In Russian, the phrase isn’t “measure twice, cut once” but “measure seven times, cut once.” To avoid the main failure traps when developing major People products, the depth of planning should be between rigorous and onerous. We need to emerge from the Scope phase with the following:

• High confidence in the team's ability to successfully ship and implement the product within existing constraints.

• High confidence that end users will engage with and perceive value in the product.

• High confidence that the product will meaningfully impact our goals, aligning with our strategy.

It’s legitimate and desirable for products to fail the Scope phase if these conditions are not met. This can be because more iteration is needed or because we have learned that developing this product is not workable now. If a product plan is not well developed here, no amount of execution will save it.

During my time at Miro, Farbod Saraf, and the Product Management team used a Product Alignment Document (PAD) to divide the Scope phase into two main parts: Opportunity/Problem Framing and Solution Framing, with a go/no-go decision gate in between. This helped teams avoid wasting time scoping a solution to a problem that, on examination, we didn’t want to solve at this time.

• Opportunity Framing asks: What problem are we trying to solve? What will be the impact of solving? How have others solved this problem?

• Solution Framing sets the parameters for the solution, maps the development process, and surfaces required resources, tradeoffs, and execution risks.

During my tenure at Miro, I customized the PAD for developing major People Products. The “Guiding Questions” help design teams determine what to work on, and the “Criteria” help them assess their progress.

Opportunity/Problem Framing:

Solution Framing:

Executive Team Touch-points

You should schedule time in the executive agenda for a brief People Product Kickoff for each feature on your roadmap. As preparation, you should have finished “Opportunity Framing” and have preliminary drafts on “Solution Framing.” The key deliverables for this time are to:

• Challenge and validate each aspect of the Solution Framing completed thus far.

• Capture insights from executives' relevant experiences.

• Align the capacity needed from leaders and functional employees in the design process.

Design sparingly

Some problems demand that we design breakthrough solutions, but most don’t. Great companies have already solved most organizational challenges well enough for us to copy, adapt, and move on quickly. This is lucky, as good design is highly resource and time-intensive, so we have to use it sparingly.

Among its remarkable achievements, SpaceX invested over $3 billion into developing reusable rockets, culminating in the Falcon 9. It was the first orbital-class rocket capable of reflight and now has launched more than any rocket in history, with 80% of launches using reused boosters. This ground-up innovation reduced launch costs by 90%, making space access more economical and reinvigorating the entire industry.

Meanwhile, in 2024, BYD became the world’s top-selling EV company—not by designing everything from scratch but by rapidly scaling existing technology. Their 41% growth overtook not just Tesla but also Honda, Nissan, and Suzuki. One approach focuses on invention, the other on refinement. People teams need both.

The most critical step in the Design phase is deciding whether to innovate like SpaceX or fast-follow like BYD. True innovation can create market-leading solutions but is resource-intensive and can risk falling into the “not shipping" if moving too slow or the “user rejection” if going too fast. Fast-following accepts proven practices and their limitations to move quickly at the risk of not being built for purpose and falling into the “doesn’t move the needle” trap. Given time and resource constraints, only a fraction of your design capacity should go toward true innovation. A team spread too thin will fail.

(Principles drawn from Elon Musk’s 2021 interview at SpaceX's Starbase facility, Everyday Astronaut).

(Principles drawn loosely from Greeven & Yip’s “Six Paths to Chinese Company Innovation”, Northeastern University)

Executive Team Touch-points

The People Product Kick-off helped us scope the solution. Now, we need to engage the executive team in validating or course-correcting our design. This is best done in two steps.

1. Early in the Design phase, one-on-one with each executive team member to test the planned solution in discussion before it is built. This should challenge key assumptions and ensure that the relevant context from the functions has surfaced through discovery.

2. Late in the Design phase, with the full executive team, to demo the full product and confirm confidence in realizing usage and business impact goals. The meeting should also align the team around the Launch timeline and roles required to play in its success.

Launch together

"Everyone’s time is already allocated." — Marc Andreessen

Most products—even beautifully designed ones—fail at launch. You might have rigorously tested with employees and validated your product’s use case, but the Launch phase is when the collective mindshare of the organization decides whether a product is worth using or just more noise.

People teams often miss this reality. Unlike external products that compete in a marketplace of solutions, People Products don’t have competitors—but they are competing for meaning and attention. Our unique ability to define mandatory practice can blind us to this challenge; compliance can get employees to perform a behavior, but it won’t drive engagement, integration, or actual impact.

Executives play the most critical role in communicating this meaning. Line managers watch them to determine priorities, and employees watch their managers. Everyone can tell when leaders read talking points they don’t understand or aren’t fully aligned with. However, when executives genuinely champion an initiative, their teams naturally integrate it into how they define their work.

If you’ve invested in executive champions during the Roadmapping, Scope, and Design phases, you’re ready to Launch together. Two disciplines have thought deeply about leveraging leadership to drive the kind of impact we seek:

• Internal Communications teaches us that how something is communicated—not just what is communicated—determines engagement. Messaging clarity, repetition, and alignment with leadership matter as much as the content itself.

• Change Management teaches us that leadership activation, short-term wins, and cultural reinforcement are required for initiatives to stick. Early visible success drives momentum, and ongoing reinforcement ensures adoption.

The model below draws on both fields to provide a structured operational guide for launching together with your executive team.

Becoming a Minimum Viable People Product Manager

Executive, people manager, business partner, talent advisor, culture builder. The Head of People’s responsibilities are broad, but adding PM for the organization brings structure to how we build each role. Instead of shielding our executive peers from the People agenda, we can adapt the PM toolkit to make it easy for them to influence and champion the products we build. But how do People Leaders learn the core skills needed to deploy this toolkit?

"There isn’t enough time on this Earth to learn everything you could about those three circles [Business, UX, and Technology], so as helpful as this diagram is, it ends up impractical... The Minimum Viable Product Manager defines a set of skills or knowledge useful to be an effective generalist product manager, one who can work on almost any problem."

• Brandon Chu, former VP of Product at Shopify, MPVM: Minimum Viable Product Manager

This is precisely the problem that People leaders face. We don’t need to be full-time PMs, but we do need the minimum viable skills and knowledge to play the PM role for our organizations. Brandom wrote a “syllabus of sorts,” and I’ve heavily modified it to fit our People Product context below.

As I’ve heavily drawn from Brandon’s excellent piece in this chapter, he deserves the closing line.

“I don’t want to trivialize learning all this stuff. It’s not easy, and it takes a lot of time, so tackle it bit by bit and enjoy what you’re learning. I hope this helps you be a little more efficient in your quest to be a great if minimally viable, product manager.”

Michael Lewis, the best-selling author of Moneyball and The Big Short, tells the story of Harold Abrahams, a British sprinter who won the gold medal at the 1924 Olympics with the help of a coach. However, the coach was not allowed to enter the stadium because “In 1924, professional coaches were taboo. They were considered a form of cheating—the steroids of the age.”

The irony, Lewis points out, is that the same elite who once shunned coaching as unsporting now embrace it wholeheartedly. “The old aristocracy is dead,” he says. “Now everybody thinks it’s good to try. Everybody competes, and the people who take their coaching most seriously are the aristocrats.” Structured guidance is no longer a dirty secret but a crucial advantage. Coaching has permeated mainstream culture—from executive coaches in corporate boardrooms to SAT prep tutors for students seeking an edge in college admissions.

Despite its growing presence, coaching remains underutilized. Even amongst CEOs, only around 40% currently have executive coaches. On the home front, only 30% of couples have engaged in shared counseling at some time. And what tiny fraction of professionals get coaching for areas as important as their personal productivity, career transitions, salary negotiations, or caring for their children or aging parents?  The few who do may be considered eccentric, but their gains are differentiating. 

“Back when coaching didn’t do very much, it didn’t matter who got it and who didn’t. But coaching is clearly getting better and better, and spreading into more areas of life, which means it matters a great deal who gets it and who does not. And the people who don’t get it are often the ones who need it most."

• Michael Lewis, Against the Rules, “The Unfair Coach”

Now, Voice AI coaches are poised to bridge these gaps, offering on-demand, nearly free coaching for all kinds of personal and professional growth. Imagine a coach available anytime—on your laptop, phone, or wearable device—ready to discuss, critique, practice, and plan. Coaching will become the interface wrapper for all the software you currently use and a menagerie of new tools built “coaching first.” 

This democratization of coaching will be the most significant transformation in how the world learns since the rise of the Web. It will be a wave that some of us ride early but ultimately raises all future generations. The Web transformed how we access information; Voice AI coaching will transform how we synthesize and apply that information, enabling us to grow, work, and thrive in ways that were once unimaginable. 

Why Her?

When we think of Voice AI, it’s hard not to conjure images from Spike Jonze’s 2013 film Her, where a soft, human-like voice (portrayed by Scarlett Johansson) guides Theodore (Joaquin Phoenix) through his work, relationships, and existential questions. In the film, voice-based AI transcends being a mere tool and becomes a trusted source of reflection and guidance. Our association with Her became even more concrete when OpenAI released Sky, a voice so similar to Johansson’s that it led to a lawsuit and the voice’s eventual retraction.

Her felt futuristic, but a decade later, Voice AI is here, with platforms like OpenAI offering a shockingly natural Voice-to-Voice experience. The technology’s ability to understand speech and deliver human-like responses has advanced remarkably. This week, OpenAI rolled out its Advanced Voice Mode to paid users, offering speech and interactivity so lifelike and customizable that OpenAI slowed its deployment to address critical risks. Their safety teams had raised concerns around the Mode’s ability to emulate any voice shared by the user convincingly, its increased ability to persuade users on political topics, and its actions that built misplaced high-trust emotional bonds with users.

To demonstrate how fluid and immersive this technology has become, I recorded a live discussion on this article with a Voice-based GPT. The natural dialogue and depth of inquiry are impressive.

This Voice-to-Voice experience builds on established capabilities of Large Language Models (LLMs) like ChatGPT and Gemini, to answer questions on myriad topics, offer cogent suggestions, and follow structured conversations. These skills are powered by vast training data, ongoing learning from user interactions, and instructions provided by custom GPT designers. Voice AI transforms these capabilities from a point solution to a new platform for optimizing the work and learning experience.

AI has brought other exciting learning modalities. Text or audio AI summaries can be integrated into all major content readers and web browsers. Google’s NotebookLM goes further, offering that “with one click, two AI hosts start up a lively deep dive discussion based on your sources. They summarize your material, make connections between topics, and banter back and forth.” 

However, individualized coaching- executive coaching, therapy, or an Oxbridge tutor- has always been the gold standard for personal development. A skilled coach can create interactivity, customization, and motivation that broadcast methods cannot replicate. Yet, the inherent one-to-one nature of traditional coaching has always been its limitation. Expert coaching is too expensive even to be used broadly, and even when in use, participants would benefit from far greater contact hours if cost were not a factor. With Voice AI coaching, these constraints are disappearing. It will offer near-perfect simulations of human coaching—on-demand, scalable to millions, and available at a fraction of the cost—unlocking personalized guidance for everyone, anytime.

The Last Revolution: Democratizing Information

The history of learning technology is a story of expanding access to information. Before the printing press, books were rare and available only to the elite. Centuries later, computers and CD-ROMs brought multimedia learning experiences into homes and classrooms. The early internet's websites, forums, and blogs offered unprecedented access to peer-to-peer knowledge sharing and niche communities. And in the new millennium, mobile apps and video learning through YouTube and massive online course sites brought access to world-class lectures to the four billion adults globally with smartphones.

This democratization of information has revolutionized learning by creating access to “omni-information”—the ability for anyone, anywhere, to access free information on virtually any topic with just a few clicks. The key beneficiaries of this wave have been self-driven learners, often called “autodidacts.” These individuals know what they want to learn and how to find it. They can synthesize knowledge from structured information sources like eBooks, wikis, and online courses and are self-motivated to learn broadly. But simple sounding skills like “knowing what to Google”, “having good reading habits”, and “following smart accounts” belie a more profound ability to access the omni-information.  It enabled them to accelerate their growth, thrive in knowledge-driven fields, and often bypass traditional education systems.

Despite the vast availability of information, self-driven learners represent only a small minority of working professionals—perhaps less than 10%. The rest of us stand at the library door, unsure where to begin, unable to turn our learning intentions into consistent practice, or feeling disconnected from the materials we can find.

The Next Revolution: Democratizing Coaching

The next breakthrough in learning technology builds on the omni-information revolution by offering access to “omni-coaching”—interactive AI coaches available for every topic, role, and growth stage. These coaches will be seamlessly integrated into software and personal practices, providing affordable, on-demand, voice-based coaching at every step of our learning and workflow cycles.

While the last wave empowered self-driven learners, the next will enable “coachable learners.” These individuals may not instinctively know what they need to learn, but a coach can guide them to set direction, synthesize personalized information, and respond to customized prompts that enhance their learning process. As a result, they will develop faster and more comprehensively across various personal and professional domains.

This new wave has the potential to engage far more learners than omni-information ever did. Whether through sports coaches, personal trainers, or mentors, many of us have experienced how a coach can motivate us to take actions we might otherwise resist. In the same way, omni-coaching will dramatically expand the scope of learning for coachable individuals. While omni-information primarily benefited a small segment of autodidacts, omni-coaching can reach nearly half of all working professionals. This would represent the largest activation of learning in human history, fundamentally reshaping organizations and even our broader culture.

What makes Voice AI a powerful coach?

The mechanics of how this new technology creates incredible learning outcomes are worth understanding. I will explore them through the lenses of three big ideas drawn from the learning-related roles in business, academia, and education that formed half of my career.

We love to be immersed in experience
From studying with Chris Dede, Harvard’s Professor of Instructional Technology, I learned that immersion is a central goal of virtual learning. Dede describes immersion as the “suspension of disbelief that one is ‘inside’ a digitally enhanced setting.” Immersion engages learners, making them active participants rather than passive consumers of content. Voice AI coaching achieves this by simulating the context of a specific natural conversation so convincingly that users respond as if interacting with a human. This elevates the learning challenge, eliciting user response, reflection, ideas, and realistic anxiety. This immersion not only brings the user back to practice but enables better transfer of learning into real-world contexts.

We want to be catered to
While immersion is a powerful precondition for learning, personalization is the true grail of education. Personalization promises to empower learners by giving them control over what, when, and how they learn, adapting experiences to their unique needs, preferences, and learning styles. While working with Teach for Australia and the University of Melbourne, I saw firsthand that even the most innovative educators could only really differentiate instruction for groups within a class. Meeting each learner exactly where they were seemed like an unrealizable ideal—a direction rather than a concrete goal.

Voice AI coaching is now poised to deliver on this promise. LLMs like ChatGPT and Gemini can generate tailored curricula on nearly infinite topics that adapt in real time to each individual's growth stage and practice, generating guidance, practice, coaching questions, and feedback customized to their styles and preferences. As discussed later in the piece, the accessibility of enterprise-grade coaching products for each use case at a near-free price point is just a matter of time.

We learn through phases that can be accelerated
David Kolb’s Experiential Learning Cycle became popular in the pragmatic world of corporate education because it aligns with the real-world challenges employees face. It suggests that knowledge is generated from experience by individuals moving through four stages: experiencing, reflecting, thinking, and acting. An unlimited Voice AI coaching resource can deepen each stage and efficiently guide learners through them.

At the experiencing stage, Voice AI provides immersive role-plays and simulations, allowing learners to engage with scenarios like giving feedback or resolving conflicts. With its ability to access work in real-time—such as through messaging apps or design tools—Voice AI offers contextual guidance that sharpens the learning process. In the reflecting phase, AI guides users through analytical questions, helping them process their actions and outcomes and prompting reflections on offline experiences. In the thinking stage, Voice AI compares new insights with past experiences, acting as a thought partner to clarify ideas and integrate them into broader frameworks. Finally, during the acting phase, it supports experimentation with new strategies, guiding users through testing and refining approaches based on their ongoing reflections and learning.

Current Applications for VoiceAI Coaches

Voice AI coaches already create substantial value in areas where they can form realistic simulations of work and learning. This technology excels in two key areas: Role-Playing Conversational Skills and Facilitating Personal Practices. The tables below outline each phase and provide relevant examples.

Applications around Role-Playing Conversational Skills work well because the voice-to-voice experience convincingly simulates human interaction. By customizing LLM instructions for specific use cases and drawing on vast training libraries, these coaches replicate realistic scenarios and offer expert feedback on various conversational skills.

Voice coaches can also currently excel at Facilitating Personal Practices by hosting structured agendas, drawing from a diverse library of prompts and practices to keep sessions fresh, thoughtfully asking and responding to user questions, and providing positive feedback to maintain users’ motivation to practice. Coaches guide users through daily and long-term practices, helping them establish routines that build toward larger goals.

Coming Applications for VoiceAI Coaches

The next phases will apply Voice AI Coaches through Enabling Interpersonal Relationships and Optimizing Individual Workflows. Cultural and technical advancements are required for Voice AI coaches to integrate successfully into interpersonal relationships. Although the core AI technology is already sufficient, gaps remain in regulatory compliance (especially in patient care) and social comfort with AI agents (in interpersonal collaboration and guest care). Nevertheless, start-ups and established companies are rapidly advancing user experience and enterprise security to make these applications a reality.

Optimizing Individual Workflows will also require further developments for widespread adoption. One of the most anticipated is giving models visual access, whereby AI coaches can see and interact with information on a user’s screen. Already, “multi-modal” capacity means that models like GPT4o and Gemini can incorporate images- like analyzing screenshots from an app. Making these views automated and prompted by AI reasoning will dramatically enhance the coach’s ability to provide contextual guidance, making it even more personalized and practical.

Advanced assessments are another critical area for growth. As AI systems become more sophisticated, they can validate their coaching effectiveness through continuous assessments, providing users with more profound insights into their progress and targeted guidance. Lastly, integration with wearables, such as smart glasses, will bring AI coaching into previously offline spaces. With the ability to perceive and understand more of the user’s environment, AI coaches can provide real-time feedback in even more contexts, from work tasks to fitness and beyond.

Try Voice AI coaching

The potential of Voice AI coaching becomes clear once you experience it. Here are some simple ways to get a sense of this technology in action:

First, try using native ChatGPT to coach you. It’s straightforward:

1. Sign in to the ChatGPT app on iPhone/Android.

2. Tap the soundbar icon (၊၊||၊) in the lower right corner.  If you don’t yet have the advanced voice model, click the headphones (🎧) icon to use regular Voice Mode.

3. You will be taken to a screen with a Blue or Black circle. 

4. Start speaking

You can ask ChatGPT anything. Here are a few prompts to try:

• “I want to practice negotiating a salary raise.”

• “Coach me on setting ambitious goals for the year.”

• “Help me learn more about [topic] and coach me through understanding it.”

Adding “custom instructions”, in the form of a few pages of prompt guidance, elevates the depth and structure of coaching the model can provide. Here’s a demo of a Feedback Coach I developed with Slate, a leading Silicon Valley coaching platform. The coach is a CustomGPT that helps managers practice giving difficult feedback, using a specific learning model and phasing to reflect on and enhance their skills.

You can also fully interact with a coach based on custom instructions through a “Personal Brand Coach” I built. This coach guides users through a seven-step personal branding process inspired by Harvard Business School's Professor Jill Avery and Rachel Greenwald.

What Will We Do with It?

We are on the brink of the largest mass activation of learners in human history. My two-year-old will enter a workforce where AI coaches are integrated into every facet of professional life. But only some will embrace the first wave. These pioneers are already experimenting with AI coaching while others wait for best practices to emerge from the cloud of new products. Still, more will be left behind—both organizations stuck in workflows of the last century and individuals lacking the coachable skills or motivation to access these tools.

For all the big-picture opportunities, the personal impact is the most intriguing. We often judge ourselves harshly for not doing things we know we’re capable of. Despite good intentions, action doesn’t always follow. Because we believe we can do it on our own, we hesitate to seek support or tools that would ensure success. Voice AI coaching can help bridge this "theory-practice gap," making it easier than ever to accomplish our goals and grow in ways that matter most. A whole lot of "I should" just became "I can."

"I feel like I can be anything with you." Theo’s line from Her captures the sense of limitless possibility that comes with deeply immersive, personalized, and efficient learning experiences. The potential for growth across any dimension is immense. But this raises a crucial question: What ideas and ideologies—conventional or radical, intentional or accidental—are embedded within these new instructors, and how will they, in turn, shape who we become?

Much has happened since I published the last piece on social crises in the workplace. The cultural loggerheads of Israel-Gaza have continued to paralyze political action within companies, even as the Security Council moves to end the war. Google fired 50 protesting employees, who then filed suit with the National Labor Relations Board for suppressing their rights. Meta removed Workplace posts from employees voicing support for people in Gaza. And Harvard announced it would no longer make public statements on issues that don’t immediately affect the “university’s core function.” 

The previous edition, When political issues bleed into the workplace, detailed the impact of corporate crises, showing the failure of organizations to keep up with the long-term social and organizational trends toward employee agency. These trends suggest that when the organizational stalemate of Israel-Palestine is superseded by an issue that has the support of a strong majority of employees, the pressure of political crises will return. Companies with high Political Risk-Factors (support for the issue, cultural expectations of employee agency, relevance of issue to to mission or key customers, limited interpersonal trust) are highly susceptible. During this apolitical summer, they would be wise to evolve their governance approaches so they can maintain the trust, unity, and speed that are the lifeblood of technology firms. 

Managing social crises in the workplace has proved extremely difficult, as it confronts a mix of ethical and identity issues, implicit norms, and leadership roles. The tensions also reinforced blind spots in how tech leaders view their organizations. We did not recognize the new qualities in our employee base and the group dynamics that were already rising before the crises. Similarly, we were unable to see solutions that would positively leverage these new dynamics. Instead, companies doubled down on their risk management toolkit and tightened management control a few notches. As a result, we missed some valuable lessons for tech company governance.

This piece will engage the problem more broadly, asking the questions: 

• how could we rethink our model of the organization in light of what we learned during the crises, 

• what new governance structures are needed to leverage these new modalities and avoid crises, and 

• how can these structures be applied to a range of different corporate political preferences: moderate, “non-political,” and progressive? 

We need new experiments that bridge the simplistic in-group, out-group binaries of our moralistic age. I hope this piece and the associated projects contribute to this urgent agenda. With Trump already contesting the legitimacy of the ‘24 election and progressives remobilizing after his felony conviction, the next political tsunami could arrive sooner than we expect.

Machine > Organism  > Community

The organizational model has evolved over the last hundred years, marking a radical increase in individual agency. The first transition, which spanned much of the 20th century, moved the model from Company as Machine to Company as Organism. The Internet Era triggered the second transition, which continues to the present day, to Company as Community.

What’s different about Company as Community?

The Community model's building block is the high-agency employee, a product of the long-term organizational social trends described earlier. This employee is highly empowered and autonomous in their work, resists narrow normative expectations, and expects to more fully integrate their personal preferences and identities into their professional lives. They seek greater alignment between their and the company’s values, including its impact beyond the workplace. This is most clear in Progressive cultures, but the same trends exist among companies seeking to attract a more Conservative talent pool- just targeting different values.

The Community model highlights that individuals bring diverse perspectives and purposes that show up in group dynamics. The alignment and fragmentation of these groups within the community bring attention to the influence of shared norms and goals, the distribution of power and decision-making, and how groups in the majority treat minority perspectives. These group dynamics can create highly cohesive, efficient, and directed companies that are resilient to external pressures or ones that are divided by conflict and paralysis. The model diagnoses both the problem and underlying causes of these political crises, showing blindspots in the previous Company as Organism model. 

When can the Company as Community model be most useful?

1. When you need resilience. Hypergrowth has compensated for many weakly held communities. Still, every company needs a second act, and the strength of the community will determine how engaged, retained, and directed your team is after growth slows, a competitor takes a punch, leadership changes, or a crisis hits.

2. When you need the next generation. High-agency individuals demand community for fulfillment and productivity. But if unmediated by appropriate structures, they can breed misdirection and mistrust, disrupting economic and cultural progress. A swing back to command-and-control approaches that have often followed only exacerbates the problem and reduces the company's ability to leverage the next generation of business and talent.

3. When you have a broken community. If there is a major division in the company a community model can help reset and rebuild. It helps us make clear the implicit norms, purpose and relationships that are at the core of our organizational strategy. A broken community is not solvable with employee engagement processes in the same way that product development can’t be addressed in customer surveys.

Subscribe now

Start-up Archipelago

So, how do we build communities that cultivate and align high-agency individuals to create excellent business and cultural outcomes? Western democracies, where the culture of agency is most advanced, are too scaled and institutionally complex to serve as an effective model. Companies are smaller, simpler, and far less democratic. They would be single-party states, with power concentrated in either a managerial class or even a single executive. Rather than contemporary society, we can find a closer analogy by going back into our history to a moment when social structures were smaller, democracy was just emerging, and the political community was first being conceptualized. 

There were around 1500 independent city-states competing in the Ancient Greek world, with citizens numbered in the high hundreds to low tens of thousands. It was a structure much more similar in scale to a start-up ecosystem. These communities had a broad spectrum of governance models, citizenry rights, and economic policies, but most were consolidated monarchies or ruled by an elite class. Protodemocracies, like Athens, were few, and their simple systems were as prone to mob rule as authoritarian states were to tyranny. 

Against this diverse political laboratory, Aristotle wrote the first empirical evaluation of social organization and its progression. In Politics, he was early to identify that it is human nature to seek community for fulfillment and belonging, the cultivation of values-based action, the exertion of moral influence in the world, and as a source of stability and change. However, he also noted that community aims were far from guaranteed and that corruption and decay were possible, regardless of the power-sharing model chosen. 

Aristotle's solution focused on a constitutional framework as the defining structure of the community. This was not just a framework for government but an expression of identity, values, and aims that aligned and educated the community. Given its expansive coverage of productivity and cultural factors, plus its flexibility across modes of governance, Aristotle’s framework provides a useful model for how a modern company can thrive with a high-agency employee base.

The Community Constitution

What’s different about Community Constituion?

• It makes norms explicit and actionable. Often, norms around power, change, conflict, and membership are implicit and can be unknowingly broken, which raises tensions.  

• It specifies purposes. Social aims are often vague, which allows overinterpretation or can make action seem arbitrary. Similarly, the community's aims are rarely specified.

• It elevates the community. Employee community is currently synonymous with events. A Community Constitution aligns existing cultural artifacts around greater community purposes, complements them with elements needed to thrive, and elevates it to the level of organizational strategy.

A Community Constitution is not a legal document like Articles of Incorporation, Corporate Bylaws, or an Employee Contract, yet if practiced, it is instilled with meaning and influence. At its core, it is a moral framework for the community, targeting aspirations for growth, collective benefit, and virtuous action. This makes it a powerful tool for reconciling the personal and ethical tensions and reactions at the heart of conflict and crises.

Implementing a Community Constitution 

The objective of a Community Constitution is to create a thriving corporate community that fulfills its purposes, where its component elements are in active practice, and with the strong support of almost all employees. There are several steps for successfully implementing such a framework.

1. Customize: Draft the Constitution for the company's unique business, culture, and governance mode, making each of the component elements explicit. This makes norms more influential and updatable, reducing tensions from members unknowingly contradicting them.

2. Activate: All community members need to be involved in incorporating the Constitution and establishing the community. Cultural influencers and leaders will need greater buy-in. Aspirational elements must be carefully implemented, closing gaps between the explicit and implicit norms. It’s essential to induct new hires into the community so they understand the framework, their roles and responsibilities, and how to drive change appropriately. 

3. Respond: When a crisis arises, norms should be strong enough that unconscious violations are rare and quickly resolved with remediation processes. Having set channels to address change proposals in light of the defined purposes should harness activist energy. If there are conscious and repeated violations, the framework serves as a robust cultural force to excise disqualifying members or an explicit set of norms to update and strengthen.

Putting the Constitution to Work

“...one might say that the best constitution is not simply the best in an absolute sense, but the one that is most suitable for a particular people." Politics, Book IV, Chapter 11.

The framework of the Community Constitution gives us the flexibility to adapt to the variety of scale-up governance modes that we see now and to plan new experiments. This final section maps dominant and novel archetypes to show how Community Constitutions could differ and how each type would help weather political crises.

Model 1: Bay Area Standard: The standard set by the rise of Silicon Valley Tech Giants and globalization of coastal culture, it is defined by moderate liberal values, strong hierarchies, speed, teamwork, and a sense that they are a force for good in the world.

Model 2: Politics of “No-Politics”: This rising alternative bans political discussions at work and seeks to undo Progressive cultural features. Coinbase and Basecamp exemplify the model that positions itself as apolitical, though it most popular in Conservative verticals and employee populations.

Model 3: A Progressive Experiment: This experiment tries to authentically incorporate progressive ideas of democracy and intersectionality into a fast-moving business to attract very progressive tech talent seeking alignment of values.

Model 4: A Classical Experiment: This fun experiment considers what Aristotle would suggest if his ideas of the ideal political community were updated for a modern tech business and from the barbarism of the ancient world to contemporary social values.

Governance for the Next Generation

As technology and economics help us more easily fulfill our basic needs, new generations seek greater actualization. Autonomy is motivating. Feeling resonant with the values you are immersed in is motivating. Yet, unmediated autonomy, which was well-inspired to close perceived values gaps, has led to political crises within organizations. We face a choice to swing back towards less empowering models of the organization or to advance with governance structures that can match the emerging ethos. 

Community is the platform upon which the high-agency talent pool can be engaged, connected, and aligned. Properly guided, a community can demonstrate resilience and change in the face of inevitable crises. Customizing a Community Constitution to the unique mission, implicit cultural norms, and economic realities provides a practical and moral framework for the company. It offers a path to business resilience and collective thriving, avoiding the pitfalls of majoritarianism and authoritarian leadership. And with the pressures of political partisanship and the transformations of AI, we will need a lot more resilience.

Thank you for reading! I would love to hear your thoughts. Feel free to reach out below, by email or on LinkedIn. Hard People Problems is a deep dive into the toughest problems in HR, delivered free every two months. Subscribe to stay in the loop and support my work.

Leave a comment

Subscribe now

HPP#2: How social currents became tech workplace crises

Tech firms have long sought to be apolitical. They perceived themselves as building a neutral infrastructure that would empower society, unfettered by outdated politics, and as having stronger shared identities than partisan divisions could break. The libertarian aspect of old California and the progressive universalism of new California made politics seem like an East Coast phenomenon, remote from where the internet was being shaped for the world.  

This illusion has evaporated over the last five years as societal and political issues have broken into the workplace, creating crises. The mechanics of employee activism and corporate response may have contributed to movements’ aims (BLM, Russia-Ukraine, Israel-Gaza), but they have also reduced trust, unity, and speed within many firms. Most recently, it is striking that in many firms, both pro-Israeli and pro-Palestinian employees feel their sense of belonging was undermined. As in an early democracy, majoritarian dynamics around issues have led to broken operating norms and undermining the rights of minorities.    

Politics is hard. When interviewing founders and people leaders about their most complex people problems, dealing with political crises at work regularly tops the list. The challenge is simultaneously practical and ethical, intimately personal and systemic, driven by deep meaning and base fear. For tech companies, which consider their shared identity, purpose, and collaborative spirit “in their DNA,” political crises can feel existential.

This edition will delve into the genesis of this challenge, tracing the roots back to the Agile methodology and the emergence of Workplace Humanization. I will give a brief history of political crises in tech, explain what factors determine if and where they will strike, and show the limitations of the current practice for managing such crises. The following edition will propose a better solution for managing crises: a cultural governance framework to sustain and evolve companies of various political orientations.

With the cultural zeitgeist shifting away from progressive activism, one could assume that the wave of political crises at work is over. However, with the strong likelihood of a second Trump term or a highly contentious election, we may be in the eye of the storm. Now is the time to learn lessons from our past practice and prepare our organizations for the coming economic, social, AI-based, and geopolitical challenges.

THE RISE OF EMPLOYEE AGENCY

There has been a 30-year trend of increasing expectations for employee agency- the capacity for employees to act independently, make decisions, and exert influence within their roles. The shift towards Agile methodologies was a significant factor in this, creating expectations that employees would self-organize much of their work and that the organization would continuously iterate with employee feedback. The Agile methodology doesn’t set limits on this feedback nor consider that employees would respond to political crises, enforce dominant group norms, or seek to change power dynamics. As a set of workflow principles raised to organizational theory, it’s unsurprising that our political spirits are outside its scope. 

The second factor driving the rise of employee agency is the rise of Workplace Humanization (WH). WH weaved together currents from Positive Psychology, the DEI movement, and the War for Talent to expand the role of employees in the workplace and what they expected from their employers. WH advocates see traditional styles as transactional or reductive and to be replaced with empathetic and holistic ones. The core principles of WH directly tie to increased employee agency.

• Encouraging authenticity, or “bring your whole self to work,” empowers employees to influence and influence from alternative perspectives. 

• Offering alignment with individual purpose and values encourages employees to be highly sensitive to their meaning and values being met and drive change or resistance around that.

• Creating a sense of belonging and community connection and mobilizing them into groups of like-minded and empowered actors.

There is a recent countertrend against WH: traditionalists who want to return greater power to leadership and have employees leave their personal needs and preferences outside of work. They propose a simple solution to the crisis challenge: “no politics at work,” which effectively bans personal or political discourse on controversial topics in the workplace. As straightforward as it sounds, defining “political” is not a bright-line rule and will be shaped by the leader's political views and perceptions of organizational needs. While it will continue to be a workable model for a subset of tech firms (especially those without significant US offices or in Conservative industries), the cultural trends I outline indicate this model will engage a decreasing share of the talent market. Given the central reliance on talent, most Western tech companies stand to gain more by adapting to the new era of employee empowerment rather than resisting it. 

“The core values of the generation are reflected in their prioritizing social activism more than previous generations and in the importance they place on working at organizations whose values align with their own, with 77% of respondents saying that it’s important. Gen Z no longer forms opinions of a company solely based on the quality of their products/services but also now on their ethics, practices and social impact.” 

• Deloitte, Understanding Generation Z in the workplace, September 2022

While it’s easy to dismiss narratives of generational changes as clickbait, some profound truths underpin them. Culture changes over time, and this change occurs mainly in the formative minds of the young. We commonly see our grandparents' generation as more dedicated, open to hierarchies, and less expressive of personal needs. This observation is supported by academic research on 340k individuals across 100 countries, suggesting newer generations are “more individualistic, more indulgent/less restrained, and less hierarchical.” These cross-cultural changes led to the rise of Workplace Humanization and suggest it will continue to expand.

Writing this piece made me realize I had inaccurate assumptions about Workplace Humanization. 

1. I thought the phenomenon was created by employee marketing, but it’s much more likely that marketing is adapting to a shift in employee preferences. 

2. I assumed WH was rooted in political ideology, as the representing organizations skewed left, but I now think it’s an underlying cultural shift that has carried a relatively obscure philosophy (intersectionality) into prominence. 

3. I thought the WH shift would be temporary. However, WH will likely stay, and thus, the standard organizational model must be updated.

In summary, Agile increases employees’ expectations of their sphere of influence, and Workplace Humanization increases the range of preferences they expect to influence. Together, they shape an employee base with greater agency and a tighter connection between what’s happening inside and outside the company. As talent is the building block of the tech sector, we need to critically consider how we manage this new quality of “human resource.” 

POLITICAL TSUNAMIS

Narrow Misses: Missing factors avoid crisis

There were external political challenges in tech before 2020, but none rose to the level of organizational crisis felt since then. In 2015, Salesforce boycotted Indiana, where it had significant offices, over a state bill restricting the rights of same-sex couples. But this move was popular with employees and founder-led, pitting Marc Benioff against then-Governor Mike Pence. Right-wing news outlets had decried that tech companies were dominated by liberal employees, who shaped policy and crowded out small conservative minorities, but this was largely ignored across firms and gained no traction. However, the election of Donald Trump in 2016 amplified the political focus in society and tech companies themselves. 

“Months before the election, there were reports of greater political tension in offices than in previous election cycles. In one survey from the American Psychological Association, 10 percent of respondents said that political discussions at work led to stress, feeling cynical, difficulty finishing work, lower work quality, and diminished productivity. Now, a new survey commissioned by BetterWorks… found that 

• 87 percent of them read political social media posts during the day, 

• and nearly 50 percent reported seeing a political conversation turning into an argument in the workplace. 

• Twenty-nine percent of respondents say they’ve been less productive since the election….

With all the political posts, it seemed like people were getting worked up, argumentative, and distracted,"

• The Atlantic, February 2017

In 2018, the #MeToo movement emerged, leading to widespread protests and calls for accountability around issues of sexual assault and misogyny. Twenty-thousand Googlers walked out to protest the company’s handling of sexual harassment and leadership accountability. Despite leading to important norm, policy, and leadership changes in the tech sector, #MeToo avoided becoming a political crisis by maintaining non-partisan support and having its cause already baked into the values and procedures of tech firms.

The rise of COVID-19, though at first non-partisan, also influenced the political climate within the tech sector. The pandemic heightened anxiety levels and increased sensitivity to leadership decisions as employees scrutinized how their companies handled the crisis. Lockdowns and social distancing measures cut off employees' external social interactions. Videoconferencing replaced physical workspaces and community gatherings, creating a fatiguing and depersonalized mode of interaction.

Perfect Wave: Confluence of factors builds crisis

In May 2020, Black Lives Matter (BLM) began the first of three political shockwaves that reverberated across society and into tech firms. BLM had support from a rare supermajority of Americans (67%), and within tech firms, nearly 9/10 employees supported the movement in protesting police violence against Black Americans. The BLM platform integrated into the left-wing policy platforms and institutions but had broad bipartisan support and a near-universal cultural moment after the murder of George Floyd. The movement’s core policy focused on dismantling power structures to reduce legal, economic, and representational racial inequities. In contrast, tech companies had a lower representation of Black employees and even fewer Black leaders. In addition to having large groups of progressive activists and a massive majority of employees in support, the conditions for political crises were built.

Americans are talking to family and friends about race and racial equality: 69%, including majorities across racial and ethnic groups, say they have done so in the last month. And 37% of those who use social networking sites say they have posted or shared content related to race or racial equality on these sites during this period.

• Pew Research, June 2020

The Russo-Ukraine war brought a second shockwave but was initially even more bipartisan in Western societies. The cultural unity to support Ukraine and Ukrainians abroad led to immediate actions for divestiture for Russia, immigration of Ukrainian and Russian talent, and support for Ukrainian employees. For most tech firms, the crisis had a low impact and followed the standard playbook that had now emerged. The level of internal political discourse increased, requiring an external statement and support for Ukrainian employees. However, it was the most significant possible political crisis for Ukrainian and Russian start-ups and Ukrainian and Russian employees. For Ukrainians who were conscripted, under attack, and/or with family members in danger- the politicization was absolute and drove advocacy in social media and within tech firms.

The final political shockwave in the trio was the Israel-Gaza war. In the week following the Oct 7 massacre and kidnapping, Israel had similar levels of public support to #BLM. The standard crisis playbook within companies was activated and accelerated by the employees who supported Israel, winning strong representation in leadership teams of many US tech companies. For Israeli start-ups, the politicization was near universal, although national programs enabled the continuation of the tech sector despite worker and funding shortages. As the death toll from the Gaza bombings rose, a rift occurred between left-wing activists and the Israeli-supportive group in tech firms. This corporate rift was widened by the rise of right-wing activism in media and universities and the cultural swing towards Palestinian sympathies as their suffering was magnified.

These three shockwaves created historic, political waves that broke across tech firms, creating crises in many. As has been recurrent in modern history, the Arab-Israeli crisis has absorbed all the energy for political change. One might assume that this third shockwave ends the phase. However, with Trump a favorite in the polls, the old traumas and tensions will likely reawaken, and a new period of political crises could begin.

A TSUNAMI WARNING SYSTEM

As shown above, not all significant political events create crises for companies and even those that do affect only a portion. So, how can we predict whether a political wave will break into a crisis in a tech company?

Subscribe now

Support for the political issue

The level of support for the political issue is the height of the waves. This determines how much pressure the event places on a company. Social majorities create mainstream news and social media pressure to react. Internal majorities create dominant group dynamics by asserting norms around political issues and empowering activists. The more other trusted companies take public action to support an issue, the greater the pressure on their peers to act. No matter how firm a company is against taking a stance on an issue, there is an amount of social, employee, and peer pressure that will make a company change. War in societies of the majority of the employees creates an example of total support.

Expectations of employee agency

Expectations of employee agency are the level of exposure to the ocean. This determines how open the company is to replicating external activist dynamics. The company’s expectations of employee agency, shaped by Agile and Workplace Humanization as described earlier, are reinforced by social expectations for individual agency. In cultures with low social expectations of agency, internal activism is restricted as hierarchies are empowered. In companies with low employer expectations of agency, norms against activism are powerfully restricting.

Business relevance

Business relevance is the presence of seawalls and defenses. It provides a bulwark to deflect potential crises and empower leadership responses. Companies with a core business or critical customer segments that oppose the issue create a countering force, making it easier to deflect or ignore the political wave. Defense tech firms, like Anduril, can select employees who want to advance US military superiority rather than protest the war. However, just having a strong industry or geographic customer base that is opposed is a sufficient counterbalance, i.e., selling to the US government or Middle-East headquartered firms.

Interpersonal trust

Trust is the strength of the infrastructure. This determines the level of interpersonal resilience in the face of pressure. Good intention is more easily assumed in companies with higher trust, and division doesn’t take root.  As social trust is easy at an early scale, this is why many companies smaller than two to three hundred people don’t experience political crises. In societies with higher political polarization, activist dynamics are more likely to be copied from society into organizations. Political crises can reduce trust and lower the threshold needed for subsequent political crises.

Looking at trends, we can expect further political crises in firms whose operations and business models enable it.

1. Social issues: Increasing populism, economic recession, AI labor replacement, and climate catastrophe could all be further triggers for political crises.

2. Dominant group dynamics: Globalization is likely to lessen the partisan breakdown of tech companies slightly, but the West Coast/NYC culture will still create dominant support for their progressive causes. 

3. Rise of agency: Expectations of agency will continue to rise as Agile within organizations and generational trends towards Workplace Humanization continue.

4. Lowered trust: Polarization continues to be strong in the West. As I will explain, these waves of political crises have diminished trust within organizations, making further crises more likely.  

MECHANICS OF POLITICAL CRISES

When a crisis unfolds, its progression often appears unpredictable and unprecedented, contributing to heightened tension and prompting a constrained response from the organization. However, after observing numerous cycles of such crises, we can delineate their mechanics into a structured process. This understanding enables us to grasp the dynamics of these intense events more clearly and allows a strategic response rather than a reactionary one.

Contemporary partisan activism on both sides seeks to silence opposition and hold those in power publicly accountable. Progressive activists are also more anti-capitalist and interested in inverting power dynamics. Company leaders often sense that these behaviors contradict operational norms and respond to the perceived threat with risk management and compliance approaches. 

IMPACT OF CURRENT PRACTICE

The current crisis and response mechanics undermine some of tech firms' most cherished attributes: their sense of trust, unity, and speed. 

The trust deficit between workers and organizations isn’t personal. It’s systemic… 80% of employees who have high levels of trust in their employers feel motivated to work, versus less than 30% of those who don’t. But less than half of workers say they trust their employer.

• Deloitte Insights, December 2023

Organizational Trust is based on perceptions of competency, alignment to shared goals, and behavior based on shared norms.  Activist employee behaviors challenge the leaders’ and dissenting employees' perceptions that activist employees are aligned with company goals and behaving according to shared norms. On the other hand, activists and supporting employees don’t feel that leaders are aligned (with what they see as the company's broader responsibilities) or competent in delivering on the changes they agree to. It’s noteworthy that all parties feel that implicit operating norms were violated. As described above, this lowered trust reduces the organization’s resilience against the next political crisis.

 “In today’s climate, people are saying, ‘I can’t work with you if you don’t share my views.’ It’s a problem HR professionals and business leaders cannot ignore… A quarter of U.S. workers (24 percent) have personally experienced political affiliation bias… compared to 12 percent of U.S. workers in 2019.”

• SHRM Research, October 2022

For decades, tech firms have stood out with a strong sense of unity, often represented in the idea of “one team.” Majoritarian dynamics mean that the politically dominant group attempts to exclude non-supportive minorities from influence in the form of barriers to collaboration or censorship. Non-supportive minorities feel silenced, unrepresented, and excluded from the organization, inherently decreasing their sense of belonging. As the political crises are formed on “wedge issues,” this group of excluded minorities is expanding.

“One of the biggest advantages that start ups have is execution speed, and you have to have this relentless operating rhythm… If you ever take your foot off the gas pedal, things will spiral out of control, snowball downwards.”

• Sam Altman, How to Start a Startup

In addition to damaging culture and relationships, a political crisis significantly impacts overall momentum. It consumes the time and emotional energy of employees, managers, and executives. The level of engagement invested is comparable to responding to a new threat by a competitor. It can take months to return to pre-crisis organizational speed.

On the other hand, the current mechanics have also resulted in some changes that contribute towards their movement’s aims. BLM has more firmly rooted norms for hiring and equitable treatment of underrepresented groups. Support for the Ukrainian diaspora across Europe has been significant. Israel mobilized talent and capital to sustain its tech sector. Given the considerable costs to the employees, cultures, and business operations, we must wonder if there could be a better way to manage crises and evolve our organizations.

IN THE NEXT EDITION

In mapping out the problem in detail over this edition, I hope it’s clear why it is hard and worth solving. So far, we’ve reviewed the social currents that led to our phase of political crises in tech firms, shared a model for predicting whether a crisis will break out in any particular company and culture, and demonstrated the limitations of our current practice for managing crises. The long arc of social trends and our near view on political events show that this problem isn’t going away. The following edition will propose the “Cultural Constitution” as a new solution, drawing lessons from the foundations of political theory. I will share examples of Constitutions for a spectrum of political orientations.

• Model 1: The Bay Area Standard

• Model 2: The Politics of No-Politics

• Model 3: A Progressive Experiment

• Model 4: A Classical Experiment

I hope you’ll join me in the next edition, which will be published in June. Thank you for reading!

Lattice's recent State of People Strategy 2024 is a challenging critique of the HR function’s ability to support scale-ups.

Key Finding #1: HR is struggling to demonstrate its impact

Key Finding #2: HR Leaders and the C-suite are disconnected

“While some HR teams are struggling to understand the impact of their work on key business outcomes, a vast majority of stakeholders outside of HR aren’t seeing it at all… Only 27% of respondents said their C-suite sees a connection between HR programs and revenue growth…(28%) sees a link to product/service quality and 30% to customer satisfaction.”

It is a difficult report card to receive. Over half of HR Leaders fell short of annual goals, with a twenty-five percentage point gap between how HR Leaders and the rest of the C-suite perceive HR’s business impact.

As with any critical feedback, there are multiple ways to react. The first is by framing the issue as one of perception rather than substance. The problem, as the report suggests, is 

 “(In HR) believing in the connection between their work and business outcomes and feeling confident in their ability to demonstrate that connection to other stakeholders.”

The article’s solutions follow suit, summarized as, We, HR Leaders, need more self-confidence in asserting our views of what’s essential. We need to use data and storytelling to gain alignment from our C-suite. Then, we can spend less time on executive alignment and more time working to drive business outcomes.

It is a comforting narrative, but that comfort should make us skeptical. It’s tempting to tell ourselves that we are right but misunderstood and that the solution is to double down on the confidence and communication skills that have helped us to get this far. But I think this is an instance of the “enemy is out there” trap.

“There is in each of us a propensity to find someone or something outside ourselves to blame when things go wrong… When we focus only on our position, we do not see how our own actions extend beyond the boundary of that position. When those actions have consequences that come back to hurt us, we misperceive these new problems as externally caused…​​ This learning disability makes it almost impossible to detect the leverage we can use in here on problems that straddle the boundary between us and out there.”

There’s another way to take on tough feedback and a much more straightforward interpretation of the mess we’re in. What if executives, including HR Leaders, perceive that we aren't making an impact simply because People initiatives are not meaningfully increasing business outcomes? And that the disconnect between HR and other executives stems from us defending our agenda rather than inviting the rest of the C-Suite to help decide the tough trade-offs needed to move the needle?

A better way to create business impact in HR

In mid-2022, I led the People function at Miro as the organization reached one and a half thousand employees with hundreds of millions in revenue. The SaaS Crash had begun; it was wartime again. On the flight to our quarterly executive offsite, I finalized a report from our last employee engagement survey. It was a ten-page memo outlining strengths, pain points, the themes that would address hot topics, and structures for distributed implementation. It was a beautiful document: comprehensive, logical, and data-informed. My CEO’s response?

“None of these things are going to move the needle for us. What is the one underlying factor that all of these initiatives will target?”

I was crestfallen, but I also knew he was right. Targeting too many metrics across employee experience, learning, teams, operations, brand, quality, etc., and running initiatives deployed across the length of a scaled HR team would get the same OK results as before. “Peanut buttering” would not significantly impact the business. And this problem wasn’t just in engagement planning. I’d consistently multiplied our HR and recruiting agendas as we scaled since I joined in 2019.

Solving as many needs as possible, in line with practice from great scale-ups, had been my mantra. It helped to 10x our headcount in 3 years, as revenue expanded significantly faster. The approach worked well during zero-to-one growth, rapidly establishing global systems where there were few. When you’re “giving out Cokes in the desert,” the impact is clear. As the organization scaled and matured, this mantra no longer fit. Moving the needle for the business would now require breaking through established behaviors and systems. I needed better ways to pick points of change, to concentrate resources on them, and to help this investment yield significant results. I needed a People strategy. 

Since leaving the executive team, I’ve tried to process this lesson and have worked with several HR Leaders on a better way to create business impact and alignment. When I was in the hot seat, I mistakenly thought I didn't need a high-level strategy, just operational rigor and good practice. I now realize I was limited in my capacity to think strategically while tending to the parallel explosions of grand scale (COVID, BLM, hyper-growth, Ukraine) and everyday scale (international expansion, VP hiring, tech stack changes, under-resourced teams). Below is the framework I wish I’d found four years ago. I hope it helps some of you who are riding that precious explosion today.

Make the most important thing the most important thing

Most HR Leaders have comprehensive operational plans outlined in documents structured to show alignment with high-level organizational goals. These plans help People team members know which of their activities to prioritize, give them specific measures to work against, and help them feel connected to company goals. As goals cascade through the teams, plans encompass most of the work in the function. We don’t significantly differentiate the level of ambition, investment, and ongoing attention across the breadth of these plans. We don’t concentrate enough resources to earn outsized benefits from winning strategies. Peanut buttering our investment across priorities creates priorities in name only.

The advantage of this practice is it doesn’t require taking any risks. We aren’t investing in anything specific. In fact, we’re following all the industry practices. As there aren’t enough resources and attention to fully invest in everything, any shortfalls can be excused by the breadth of parallel work and the lack of resources for any specific initiative. But this safe path also leads to the impact and alignment mess that we are in, so a different approach is needed.

If we want outsized impact from a few areas, we need to make the most important things the most important things. This means

• selecting focuses where investment and attention have the best chance of impacting business outcomes,

• investing in them through big goals, talent, and team operations,

• adding design and management attention to help iterate and increase the chance of success,

• deprioritizing investment and opting for incremental efficiency gains outside the focus.

What are the most important things?

Strategy is a guide for determining the most important things. Michael Porter, the godfather of corporate strategy, said,

“The essence of strategy is choosing what not to do. Without tradeoffs, there would be no need for choice... more of one thing necessitates less of another."

A good strategy must make choices, prioritize, and then inform cascading choices and prioritization throughout the function. Chosen areas must correlate highly with business success (measured by outcomes like productivity, product quality, cost, revenue, customer satisfaction, etc). They also need to be few in number, clear, and debatable. For scale-ups, I propose four focuses to create a People strategy.

Subscribe

This set might only seem controversial once we note the tradeoffs in selecting these focuses. The questions target specific outcomes: speed of teams, differentiation of recruiting, employee motivation and alignment, and increasing people operations efficiency. They don’t focus on employee satisfaction, experience, cost, capability, compliance, retention, brand, inclusion, or other vital factors. These should still be part of tracking and effort but not prioritized for our strategies focus as they aren’t as closely linked to medium-term business success in scaling tech firms. “More of one thing necessitates less of another.” You might disagree with the selection and find another set that better links to business impact in your organization, but you will still have to trade off against the rest. Here’s my view on why these four are critical for HR to impact the business's success. 

1. How will you help teams across the company move faster/better?

Reed Hoffman described the hyperspeed scaling of start-ups “as a marathon of sprints.” The speed - and the speed of change - is incomparable to anything in industry. You need to build a lot, in parallel and right now. If you do, we’ll see if it works, and you win the chance to build even more. If you don’t, then thanks for playing. And maybe this game ain’t for you.

Successful start-ups all start fast. Good ones sustain the pace until they hit Dunbar’s Number of 150 people, where social group dynamics break and they need to systematize much more. Here, even they start to slow. At first, change is barely noticeable because they are moving so fast, but over time, sand in the gears of decision-making and communication begins to show, and the sense of momentum drops. By 500 employees, process friction, coordination costs, and enterprise maturity can quarter the speed of the earlier days. And by thousands, only the generational start-ups are still in top gear. Once momentum is lost, it is tough to get back. Executive teams should treat organizational speed as the most precious commodity, and People Leaders should ensure that they do no harm before they do good.

“Teams” are the focus instead of “individuals” because team performance is so sensitive to process quality within and around teams. Most start-ups are now team-focused in goal setting, design, and implementation. “Moving faster/better” is a proxy for team effectiveness. “Team Speed” is the most straightforward to understand and action, but other aspects that capture the essence of team effectiveness are also workable. Aligning on a choice gives a north star for your People team, a lens to learn from high and low team performance and a yardstick for iterating initiatives. 

Your answer to “How will you help teams across the company move faster/better?” is your Organizational Agility strategy.

2. How will you differentiate to compete for the talent you need?

No one needs convincing that companies benefit from great talent. This consensus frames our challenge. The top half of start-ups are trying to get the top 10% of start-up talent. The market is inherently fierce, and you must pick your competitive strategy. The most significant factor, just how fast your business is growing (in revenue, users, funding, headcount, etc), is already your company’s biggest focus. But the locations you are competing in, what specific great talent you’re competing for, how much money and organizational focus you’ll spend fighting for it, your remote/in-person working model, and what significant intangibles you can bring to the fight are all open variables to play with.

Competition means you are not alone. Other good and great start-ups are dialing in their variables also. If you’re a 7 and they are an 8, they will win, unless you saved the point to spend on a more impactful area. But the best competition is not to compete. Blue Ocean Strategy suggests

“Competing in overcrowded industries is no way to sustain high performance. The real opportunity is to create blue oceans of uncontested market space.”

Before COVID, fully remote start-ups (like Elastic and Gitlab) deployed this effectively. There were few opportunities for skilled software developers in regional Switzerland, Kentucky, or three hours outside San Francisco. Similarly, ten years ago, the first US scale-up entrants into Poland, Ukraine, and Russia chose dynamics where their product (USD tech jobs) was strongly competed for by some of the most brilliant engineers in the world. These blue oceans seem obvious now, but they weren’t then. Now, it could just as likely be an online community adjacent to your product as an immigration opportunity from a challenging location. If you can find a blue ocean, invest in it and don’t compete. Otherwise, you must recognize and exploit your combined strengths against other good scale-ups.

Your answer to “How will you differentiate to compete for the talent you need?” is your Talent Competition strategy. 

3. How will you build a culture that motivates and aligns employees?

“Culture Amp believes in the importance of putting culture first. Companies that share this belief (which we call culture first companies) recognize that if you take care of your people and your culture, then the customer experience and profits will take care of themselves… in our experience working with over 6,500 companies, we have found six key elements that successful culture first companies have in common. Transparency, open communication, inclusivity, accountability, values-based decision-making, leading by example.”

These are all fine factors that I would like to see in my employer, but do they drive business outcomes? Culture Amp’s annual conference featured a panel on “The business outcome of putting culture first,” with prominent examples of organizations that successfully prioritized the Culture First principles. The profiled companies included WeWork (‘47 Billion to bankruptcy’),  Casper (acquired by PE as losses spiraled), and Patreon (‘Mismanagement thwarts a Pandemic star’). I don’t mean any slight against these firms or suggest these culture leaders weren’t delivering a great agenda in line with the six characteristics above. The example indicates that focusing on these factors is insufficient for driving business success. Ben Horowitz said it best ten years ago.

“The world is full of bankrupt companies with world-class cultures. Culture does not make a company. So, why bother with culture at all?

• It matters to the extent that it can help you (build a 10x better product, win the market)…

• As your company grows, culture can help you preserve your key values, make your company a better place to work, and help it perform better in the future.”

So, what cultural factors are most closely linked to business impact? A culture’s flavor is shaped by selecting between focuses like customers, efficiency, innovation, quality, impact, compliance, collaboration, empowerment, well-being, and more. The strength of the culture is determined by the degree to which its talent and interpersonal processes represent this focus. There are many viable, strong flavors, but those that most correlate with impact all contain high employee “motivation” and “alignment”.

• Motivation is the commitment, energy, and creativity individuals bring to their work. Motivated employees make new things happen; demotivated ones, at best, keep things going. 

• Alignment is the degree to which individuals direct their work towards common goals and how they weigh these goals above their individual interests. Highly aligned companies put more shots on goal and can pivot when external circumstances change. Poorly aligned ones find their efforts dissipated across the many motivations of internal or external politics.

Motivation and alignment can form a positive flywheel, which we see most easily in quality early-stage start-ups. Start-up talent is self-selected for motivation and is easily screened for alignment by founders. The shared and mission-based sense of success drives intrinsic motivation. Small group social dynamics mean the whole team feels seen, creating powerful aligning forces. Team members who are not aligned are easily recognized and parted with, reinforcing the shared norms of the continuing group. The flywheel creates a powerful cultural force with the sense that we are running fast, together.

We need to build this positive cycle at each growth stage. At first, this is just the founding team’s personal characteristics and working styles. With scale, it includes shared practices and rituals that shape group dynamics. With further scale, it systematizes into processes, symbols, and rules that guide choices and behaviors for the many. The practices that meant so much to a group of 50 might be stale or not deliverable to a group of 500. 

Your answer to “How will you build a culture that motivates and aligns employees?” is your Culture Engagement strategy.

4. How will the People team deliver these strategies with increasing efficiency as you scale?

Getting clear and invested in your Organizational Agility, Talent Competition, and Culture Engagement strategies is half the battle. Then, you must determine how to deliver the big bets that drive them. I think of a hierarchy of delivery for People teams:

1. Hustle: Being able to deliver

2. +Practice: Being able to deliver in line with good practice

3. ++Design: Being able to deliver customized practice that yields outsized impact

If your teams aren’t producing, it doesn’t matter if you have best practices or great user discovery. If you deliver quickly and in line with industry practices, you are already at a competitive standard. But for the business impact we seek, we will need all this, customized to the organization's unique needs, and iterated until successful.

Agile is the least bad way to build new things that actually work, at scale. I’ve led agile HR teams at Salesforce and Miro, and too often, I’ve added #3 before securing #2, resulting in deep inefficiency. Design without delivery is a daydream. Delivery without design is a nightmare. But agile, loosely defined, is the only way I’ve seen #1, #2, and #3 come together. Whatever approach you choose, you will need to dial in your talent, team structures, design methodology, tracking, and iterative attention just to deliver the first three strategies.

Doing so with “increasing efficiency as you scale” seems either obvious or near impossible, largely dependent on whether you’ve run an HR function. Recruiting headcount grows close to linearly with organizational growth. Shifting from generalists to specialists across the department brings scale but adds coordination costs and loss of direct engagement that can quickly eat efficiency gains. Part of the answer is to ensure we are building our priority people teams with the speed, talent quality, and motivated alignment we are trying to bring to the rest of the organization.

The other element is transformative technology. Until last year, the rule in HRIS followed Horowitz’s guidance that there are no silver bullets, only lead ones. With generative AI, that might have changed. The best teams will be highly competent in selecting, deploying, and coaching around these tools and will be far leaner.

Your answer to “How will the People team deliver these strategies with increasing efficiency as you scale?” is your Operational Delivery strategy.

Subscribe

How do you make things important?

Now that we’ve selected the most important things, we need to make them important by concentrating ambition, resources, and attention on them. This concentration requires an antidote to the “peanut buttering” approach that has spread resources evenly on as big an agenda as possible. The antidote is the 80/20 rule.

“The Pareto principle (also known as the 80/20 rule) is a phenomenon that states that roughly 80% of outcomes come from 20% of causes… The goal is not to minimize the amount of effort, but to focus your effort on a specific portion of work to create a bigger impact.”

Good strategy helps us apply the 80/20 rule to divide our agenda into “Innovation” and “Foundation” categories and to concentrate our investment on potential winners. Making tradeoffs is emotionally tricky work. Saying “yes” to investing in an area requires a vulnerability that you might fail. Saying “no” requires the vulnerability of rejection. A good strategy helps depersonalize these decisions and makes this challenging work easier.

Some Innovation Work will create an outsized impact as intended, and others will fail. Close tracking and leader participation in the discovery and testing phases will bend the curve to the positive. Nonetheless, if leaders take broad accountability when we miss, they can require that lessons are learned to improve future cycles. The winners will have a transformative impact, showing the People team and broader culture that change is possible. Recognition for teams should be similarly outsized when they are successful. Both accountability and recognition ensure that the most important things stay the most important things.

Where to from here?

Are you ready for this? Aligning an executive team on a People strategy and shaping investment around it is an advanced practice. We can use this piece's Operational Delivery competency model to assess our planning maturity.

1. Hustle: Making operational plans in a timely and detailed way.

2. +Practice: Plans reflect good practice for peers.

3. ++Design: Plans reflect strategy, with concentrated investment in breakthroughs.

If you haven’t locked in #1 or# 2, you can likely make efficient gains there. If you are already up to standard, you are ready for #3.

A central theme of this piece is that “focus is a trade-off.” I hope the investment in reading it pays off in real breakthroughs for your organization. If I can be helpful in this work, please contact me at aj@ajosephson.com for advisory or speaking inquiries. Thank you for reading! I look forward to your thoughts on this piece and ideas for what Hard People Problems are most important to solve.

Leave a comment