Company Brain and context

What is a playbook? Definition and examples

Also: Runbook, standard operating procedure

In short

A playbook describes a recurring task so that people and AI agents run it the same way: with its knowledge, steps and approvals. The result then no longer depends on who happens to do the work.

Why it matters: Good ways of working become repeatable, even when an AI agent does the work.

  1. 01

    Trigger

    A recurring task comes up

  2. 02

    Load knowledge

    Current decisions, rules, sources

  3. 03

    Run the steps

    Person or AI agent, same way

  4. 04

    Approve

    A person checks what matters

How a playbook runs

Why does Playbook matter for your company?

In many companies the best way of doing something lives in a few people's heads. Every offer review, quarterly review and customer escalation runs a little differently. For people that is tiring; for AI agents it is a problem, because without clear instructions they guess.

A playbook records which knowledge a task needs, which steps follow and where a person approves. People and AI agents work from it, and improvements flow back into the playbook instead of into one person's habits.

Example: Checking every offer against the strategy

After the decision “Strategy 2027: focus on the DACH mid-market, no US expansion”, every sales lead reviews offers differently. One asks about the market, one about the margin, one not at all. An AI assistant keeps drafting offers for US customers because nobody gave it the new rule.

With an “offer review” playbook the routine is recorded once: load the strategy and price list, check market and margin, get approval above a threshold. People and AI agents follow the same steps, and leadership sees which offers deviate from the decision.

3 common mistakes to avoid

  • Writing a document nobody uses

    A playbook is not a slide deck for the archive. It is read every time the task runs, by people and AI agents alike.

  • Leaving out the knowledge

    Steps alone are not enough. Without the current decisions, prices and rules an AI agent follows the instructions neatly, but from the wrong state.

  • Building in no approvals

    Where a mistake gets expensive, a person needs to approve. A playbook says where that is instead of leaving it to chance.

Step by step

  1. 1

    Pick one recurring task

    Choose a task that comes up often and runs differently every time today.

  2. 2

    Record the best way

    Ask the person who does it best and write the steps down briefly.

  3. 3

    Add knowledge and approvals

    Link the sources it needs and decide where a person approves.

  4. 4

    Test with people and AI agents

    Let both work from it and compare the results.

  5. 5

    Improve after every run

    Whatever does not fit, change in the playbook, not in a note beside it.

Bring one important decision.

In 30 minutes with Marc or Simon we show where your work drifts from the decision today.

Meet the Company Brain

Read on in the PM Lab

Articles that go deeper.

Related terms

Back to the glossary