Every organization has two org charts.
The first one everyone can see — boxes, lines, titles, reporting relationships, org design decks refreshed every eighteen months. The second one is quieter. It is the map of who actually makes decisions when AI enters the workflow, and it is rarely drawn.
AI in the Org Chart calls this the Second Map.
What it is
The Second Map is the working diagram of every AI-influenced decision in an organization. It shows:
- The model — the AI system in use.
- The workflow — the piece of business the model touches.
- The translator — the person or team that frames the model’s output before a human acts on it.
- The trusting party — the leader who takes the framed output as input to a decision.
- The Human Point of Consequence — the person who carries the meaning of the result.
None of these roles appear on the visible org chart. All of them shape what happens.
Why every organization has one
The moment an AI system influences work, a Second Map begins to form. It does not wait for the organization to authorize it. A shortlist model produces a slate → someone chooses which candidates to interview → a hiring manager decides. That is a Second Map — three roles, one of which is not human, and none of which is drawn on the chart.
The organization can name that map, or it can pretend it doesn’t exist. Both choices carry consequence.
What drawing it does
Naming the Second Map does three things at once:
- It exposes the translator. The person who reframes the model’s output is often the most powerful person in the sequence — and the least visible. Naming them creates the possibility of judgment.
- It makes trust legible. “I trusted the model” is not a plan. “I trusted the translator’s summary of the model, which I know is calibrated for these five inputs” is a plan.
- It restores the ability to intervene. You cannot stop a system you cannot see. Drawing the map is the precondition of governance.
How to draw it
For any consequential AI-influenced workflow, produce a one-page diagram with:
- Every role — human and system — in the sequence.
- The direction of information flow.
- The named Human Point of Consequence.
- The Space of Autonomy inside which the model may act without human intervention.
- The Challenge Path by which the decision can be questioned.
This is not a compliance artifact. It is a working document, kept current by the team that owns the workflow, and reviewed by the leader who carries the consequence.
The point
The visible org chart is the story an organization tells itself about who decides. The Second Map is what actually happens. Governing AI means closing the distance between the two.