2026-05-14
May 14, 2026
Influence at staff+ and director+ is not a meeting skill. It is an artifact skill — and most senior tech leaders are still optimizing the wrong loop.
The most common career limiter at staff+/director+ is not technical depth, strategy, or even stamina. It is the quiet belief that influence is something you do in real time. Open-mic charisma. Reading the room. A well-timed counter-argument when the VP is about to commit to the wrong thing. Most senior tech leaders invest, year after year, in becoming a little sharper in the moment — and the influence never compounds.
The leaders who scale do something different. They do not persuade. They pre-commit. By the time the meeting opens, the decision is already made — not because they railroaded it, but because their position has been visible in artifacts (an RFC, a working-backwards memo, a written weekly, a Slack thread tied to a doc) long enough that the room has metabolized it. The meeting confirms what the artifact decided.
This is not new — Amazon, Stripe, GitLab, and a generation of writing-first orgs have practiced it for years. What is new in 2026 is that the mechanic has become structural. AI agents are drafting half the analysis. Async-first teams have erased the synchronous room as the decision venue. Half the people who matter to your program will read your position from a Pragmatic Engineer-style summary, not from your face. Charisma is now a depreciating asset. Artifacts compound.
Most senior TPMs and tech leaders rise on the same set of skills: articulate in the moment, fast on their feet, emotionally smart, good at reading silence. These get you to staff or director. Past that, they stop scaling. There are too many rooms, too many timezones, too many half-attended threads, too many stakeholders who will only ever encounter your position via a doc someone else summarized. You cannot be in every room. The pre-commit can.
The hidden mechanic that the leaders who scale are running is roughly this: they have a small, repeating set of document genres — RFCs, ADRs, written weeklies, working-backwards memos, decision logs — and the discipline is that those documents arrive before the meeting that would have decided things otherwise. The artifact is not a record of the decision. The artifact is the decision. The meeting is a check.
The frame that makes this stick is the Trust Equation (Maister, Green, Galford, 2000): Trust = (Credibility + Reliability + Intimacy) ÷ Self-Orientation. Most leaders, when their influence stalls, attack the numerator — they try to look more credible, deliver more reliably, build more 1:1 rapport. Useful, but bounded. The denominator is the silent killer. In every reorg, the leaders whose influence collapses overnight are the ones whose pre-commits read as positioning rather than service. The artifact has to argue for the org's outcome, not yours. If the doc can be read either way, your trust depreciates in the silent reading, long before you ever notice.
This is also why Camille Fournier's critique of Manager READMEs lands the way it does. A README looks like a pre-commit, but it is a declaration of identity, not a position on a problem. Worse, when behavior contradicts it — and it will, occasionally — credibility falls faster than if you had never published. Fournier's framing of "the Dunning-Kruger of self-awareness" is exactly the same trap: a leader believes the README is a vulnerability artifact, but it is read as marketing. Pre-commit on decisions, not on identity. The Wardley/Amazon writing tradition pre-commits on a position about a problem — testable, falsifiable, time-stamped. The README tradition pre-commits on a self-portrait. The first compounds. The second is brittle.
The 2026 angle sharpens this further. As agentic coding tools take more of execution, what is left for tech leaders is judgment — and judgment is only legible when it is written down. Charity Majors has been arguing that the scarce skill of this era is clarity: specifying constraints, invariants, and what "good" looks like. That is another way of saying that the leaders who can write the pre-commit win the next era. Leaders whose only mode is presenting the pre-commit in the room are quietly being deleted from the rooms that matter, because the rooms themselves are being deleted.
There is also a darker implication for the manager track. Andy Grove's old formulation — a manager's output is the output of their organization plus the output of the neighboring organizations under their influence — was always a definition of leverage. In 2026, the second term is the only one that still grows. The first term is being absorbed by tooling. Which means the artifact muscle is the leverage muscle. If you are not writing the pre-commits, you are not generating output, no matter how many meetings you are crushing.
A few tactical implications for senior TPMs and tech leaders this week:
The framing to internalize: influence at staff+ is not a meeting skill. It is an artifact skill, run as a small, repeating system, in service of the org rather than the self. Get the system right and you stop having to be in every room.
Try this week. Pick one decision your team has been re-litigating in meetings for two or more weeks. Write a 1–2 page pre-commit memo: the decision you propose, the alternative you considered, the trade-offs, the date you'll move forward absent objection. Circulate 48 hours before the next meeting. Watch what happens to the meeting — and to your own meeting-prep anxiety.
What it is. A four-variable diagnostic for the trust a specific stakeholder places in you. From The Trusted Advisor: Trust = (Credibility + Reliability + Intimacy) ÷ Self-Orientation. Credibility is words and expertise. Reliability is the consistency of action over time. Intimacy is the psychological safety others feel sharing with you. Self-Orientation is the perception that you are acting in your own interest — and because it is the denominator, it divides everything else.
When to use it. When influence has stalled with a specific stakeholder. Before a reorg-adjacent conversation. When prepping for an executive who treats you as a contractor rather than a partner. When a team is technically right and politically losing. When you have just lost a budget fight you thought you would win.
How to run it:
When NOT to use it. When the gap is a straightforward capability problem, not a trust problem. Don't pathologize a missing skill — sometimes the executive doesn't trust your judgment because you have not yet demonstrated the judgment, and the answer is reps, not a Trust Equation worksheet.
Worked example. A director loses a platform-budget fight three quarters running. Every memo subtly argues why their org should be funded — measured impact, hiring plans, attrition risk. The director rewrites the next memo to argue what the company should fund, including a scenario where the recommendation points to a peer org rather than their own. They lose nothing — the next funding cycle returns more budget than the previous three combined. Self-Orientation moved. The numerator was never the problem.
The State of Engineering Management in 2026 — Jellyfish surveyed 600+ engineering leaders: 64% report 25%+ velocity gains from AI, but most are still measuring activity rather than judgment. Read it as a leading indicator of the metric you'll be defending in your next planning cycle.
What happens when engineering teams reorganize around AI agents — Smaller teams, larger scopes, and a quiet collapse of the middle of the engineering ladder. The org-design conversation senior tech leaders should be running this quarter, not next year.
AI Tooling for Software Engineers in 2026 — Gergely Orosz's annual tooling read. Skim it as a market map so you can pre-commit, in writing, to a position on which categories your org standardizes on before the AI-tools sprawl becomes a governance problem.
"A manager's output is the output of the organization under their supervision, plus the output of the neighboring organizations under their influence."
— Andy Grove, High Output Management (1983)
Don't miss what's next. Subscribe to Critical Path: