Start here
Playbook library

Playbooks

Step-by-step guides for the work of AI transformation. Every playbook states what you'll leave with, how long it takes, and the trap — one specific way teams fail at that step. Sorted by stage and by discipline.

14 published 4 in progress Reviewed by practitioners Updated

Every playbook, the same five things

  • What you'll leave with One artifact you can use — not a summary.
  • How long it takes A realistic timebox, including decisions.
  • The trap One specific way teams fail at this step.
  • Owner The role who has to hand this to someone else.
  • What must already exist The prerequisites, so you don't skip ahead.

Browse all playbooks

Filter by stage or discipline. Combined filters work — "governance for Autonomate" is one click away.

Stage

Discipline

14 of 14 playbooks

  • Strategy All stages

    Building an AI strategy tied to business outcomes

    Connect AI investment to the goals your board already tracks.

    Time
    12 min
    Leaves you with
    One-page strategy

    Trap: Writing a strategy in isolation from what the business actually measures.

    Updated Read

  • Strategy Pilot

    Choosing what to pilot first

    Pick the workflow to pilot that will fund the next ten.

    Time
    9 min
    Leaves you with
    Ranked shortlist

    Trap: Picking the most exciting workflow instead of the most measurable one.

    Updated Read

  • Strategy Pilot → Transform

    Writing the business case for transformation

    Frame AI investment in terms finance and the board actually count.

    Time
    11 min
    Leaves you with
    Baseline + ceiling case

    Trap: Building a case on adoption metrics nobody in finance believes.

    Updated Read

  • Design Pilot

    Mapping a workflow before automating it

    Capture data in, transforms, integrations, and outputs — before you touch a tool.

    Time
    13 min
    Leaves you with
    Workflow map

    Trap: Automating a workflow whose data inputs change every week.

    Updated Read

  • Design Expand

    Designing the operator role

    Define a role that runs the workflow in production — not just uses it.

    Time
    10 min
    Leaves you with
    Role charter

    Trap: Writing a role that reports into a function with no budget for it.

    Updated Read

  • Design Executize

    Building the coordination layer

    Design a controller agent that can see, improve, and extend workflows.

    Time
    15 min
    Leaves you with
    Coordination architecture

    Trap: Building the brain before the registry, evals, and decision log exist.

    Updated Read

  • Adoption Expand

    Driving adoption past the enthusiasts

    Move from power users to operators of the workflow.

    Time
    9 min
    Leaves you with
    Adoption plan by team

    Trap: Confusing adoption with licenses shipped.

    Updated Read

  • Adoption Transform

    Change management when workflows get redesigned

    Manage the human side when old roles retire and new ones appear.

    Time
    12 min
    Leaves you with
    Change plan

    Trap: Treating it as communications instead of role design.

    Updated Read

  • Governance Pilot

    Setting HITL thresholds

    Decide when the workflow runs itself and when a person reviews.

    Time
    10 min
    Leaves you with
    Tiered HITL policy

    Trap: Setting thresholds once and never reviewing them again.

    Updated Read

  • Governance Executize

    Building the decision log

    Record every extension and improvement — auditable, revertible.

    Time
    11 min
    Leaves you with
    Decision-log schema

    Trap: Logging outcomes without the reasoning that produced them.

    Updated Read

  • Governance Autonomate

    AI governance for autonomous workflows

    Tiered review sized to the autonomy tier of each workflow.

    Time
    14 min
    Leaves you with
    Tiered review model

    Trap: Applying uniform review to L0 and L4 workflows.

    Updated Read

  • Measurement Pilot

    Setting a baseline before you pilot

    Set the baseline, then the metric, then the target. Not the other way.

    Time
    8 min
    Leaves you with
    Baseline pack

    Trap: Chasing a target without a baseline to prove it moved.

    Updated Read

  • Measurement Pilot → Transform

    Measuring AI ROI and productivity gains

    The math that holds up in a budget review — including the HITL numbers most teams skip.

    Time
    11 min
    Leaves you with
    Measurement plan

    Trap: Reporting time saved without the review burden it added.

    Updated Read

  • Measurement Executize

    Evaluating agent quality

    Build an eval harness for the agents inside your workflows.

    Time
    13 min
    Leaves you with
    Eval harness

    Trap: Testing once at launch and never again.

    Updated Read

What we're writing next

  • Design · Transform

    Designing a new workflow from scratch

    How to build a workflow that wasn't possible before AI — not automate an existing one.

  • Measurement · Autonomate

    Tracking autonomy-tier incidents

    What to measure when the workflow runs itself, and what to alert on.

  • Strategy · Executize

    What an executive intelligence agent should decide

    The decisions the coordination layer is permitted to make — and the ones it isn't.

  • Adoption · Executize

    Building an AI-savvy transformation office

    What the central team's job becomes when functions own their own workflows.

How these playbooks are made

Every playbook is written by a practitioner who has run the workflow — reviewed by someone who has run it at a different organization, in a different sector. Sources cited. Methodology published where relevant.

  • Practitioner-authored, second-practitioner-reviewed
  • Every trap is observed, not hypothesized
  • Commercial relationships disclosed on the page they affect
  • Corrections logged publicly — see the log

Search and jump to

↑↓ to move Enter to search everything