AI Operating Charter

The Charter fits on one page on purpose. If a layer needs more than a paragraph, the layer is not yet thought through.

​

Genesis

Amy Edmondson (1999, Administrative Science Quarterly) named the psychological-safety construct that the first layer rests on. Chris Argyris (1991, Harvard Business Review) and Peter Senge (1990, The Fifth Discipline) supplied the team-learning ground that the knowledge-institutionalisation layer codifies. Andrew Grove (1983, High Output Management) and John Doerr (2018, Measure What Matters) supplied the OKR mechanics that the sixth layer borrows from. Phanish Puranam (2024, Re-Humanize) reads the question of human work in the GenAI workplace and gives the team-operating-model framing the Charter sits inside.

​

Why it matters now

AI use in 2026 is still mostly individually heroic. The leader who has worked their way to fluency carries a practice the team has not absorbed. Niederhoffer and Hancock (fall 2025, BetterUp and Stanford) measured workslop, AI-produced output handed off without thought, as a team-level cost. Individual delegation cycles, relationship choices and mode choices do not institutionalise on their own. Without a shared operating model, productive AI use evaporates when the leader’s attention moves elsewhere, and the agentic era arrives without a frame to hold it.

​

The moves

Three moves stand up the Charter.

​

Draft. The leader writes a first pass across the six layers: one paragraph per layer, one measurable commitment per layer. No more than a page.

​

Ratify. The team meets to debate, amend and sign. The act of debating the psychological-safety layer is itself the first instance of psychological safety at work in the calendar. Edits are expected. A Charter that ships unchanged from the draft has not been ratified.

​

Run. The Charter goes on the team page, gets read at the start of the monthly review, and amends when the work shows the layer no longer fits. A Charter that never amends has stopped being a working document.

​

The discipline

Three habits embed the Charter as operating practice.

​

A monthly Charter review at the start of the team’s standing operations meeting: which commitment is the team meeting, which is it not, what is the amendment.

​

A failure-surfacing slot, fifteen minutes once a fortnight where team members name AI work that went badly. This is how the psychological-safety layer shows up in practice.

​

A capture rollup from individual The AI Investment Loop practice to team knowledge: prompts that worked, outputs that misled, evaluations that held. Without rollup, individual capture stays individual.

​

The discipline is to keep the Charter alive: drafted, ratified, run and amended. A Charter that lives on a slide deck has not landed.

​

‘Productive AI use is a property of the team, not the leader. The Charter is what makes it stick.’ 

 

 

Viren Lall, Managing Director,

ChangeSchool LDN (2026).

virenlall.com/ai-operating-charter

​

​

​