Activator Labs 101 participant workbook

# Activators

# Enablement

# Champions

Choose one recurring team or functional workflow and prepare the right next decision to design, develop, or move it into responsible operation.

July 8, 2026 · Last updated on July 17, 2026

This workbook is for Agent Activators: the people who own or materially shape a recurring workflow for a team or function. Use it during or after Activator Labs 101 to pressure-test one workflow, then start at the first important unfinished decision in the Design, Develop, or Operationalize method. You are not expected to complete every section during the session. Capture one concrete decision, question, or rough draft and name the person who can validate it or act on it next.

Before Activator Labs 101

Bring one real, recurring, multi-step workflow your team or function already performs and that you know well. A workflow is the connected set of steps, decisions, handoffs, and outputs used to produce a result. During the lab, you will decide whether to keep it as scoped, narrow it to a manageable first path, or choose a different workflow.

Choose a valuable, feasible first workflow

	Description	Notes / Rating
Workflow	Name the recurring, multi-step work and the team or users it supports.
Impact / Value	Consider frequency, repeatability, reach, visibility, and the consequence of improving—or not improving—the result.	High / Medium / Low
Complexity / Effort	Consider process and ownership clarity, dependencies, data and tool access, approvals, sensitivity, and exceptions.	High / Medium / Low
Starting-point decision	1. Good starting point (high value / manageable effort) 2. Valuable but complex (narrow it or plan with sponsorship) 3. Optional improvement (defer if higher-value work is available) 4. Not a priority (low value / high effort)
Decision	Keep scope, narrow it, or choose another workflow.	Keep / Narrow / Change

READINESS SIGNAL: Missing or inconsistent inputs, an unclear process or owner, many dependencies or integrations, sensitive information, heavy approval needs, or numerous exceptions increase complexity. Narrow or standardize the work before adding AI.

Design: Define the current process, intended outcome, and boundaries

Start with the work as it happens today, not with a tool. Map the process, define the intended outcome and first scope, and make explicit which steps AI may support versus which decisions people must retain. Define when the workflow must stop, ask for information, or escalate to a person.

Decision	Notes
Intended outcome: What meaningful team result should be measurably better, without prescribing an AI feature?
Start + end: What triggers the work, and what counts as complete?
Inputs + rules: What information, criteria, or policies shape the work? Which approved sources provide them?
Steps, decisions, and handoffs: What happens today, in what order, and who is involved?
Output: What must be produced, decided, or changed?
Friction + exceptions: Where does work wait, loop back, require judgment, vary, or create avoidable rework?
First scope: What is included in the first end-to-end path, and what is intentionally outside it?
AI and human boundary: Which bounded steps may AI complete or prepare, which decisions must people retain, and when must the workflow stop, ask, or escalate?

Design: Questions to consider

What would the person accountable for the workflow recognize as a better result?

Which decisions require judgment, authority, accountability, approval, or sensitive context?

What approved information may the workflow use, and what must it never guess?

What should cause the workflow to stop, ask for information, or escalate to a person?

Develop: Turn the design into requirements, build, and test

Translate the agreed workflow design into concrete requirements a builder can implement and a reviewer can test. Define required and prohibited AI behavior, approved information and access, human review and escalation, and clear pass conditions. Then build—or coordinate the build—and test representative real-work cases.

Decision	Notes
First useful end-to-end version: What is the narrowest path that creates a meaningful result and tests the riskiest assumption?
Required + prohibited AI behavior and approved output or action
Required information and approved sources
Approved tools, connections, and minimum permissions
Human review, approval, and escalation
Stop, ask, or escalate condition: When must the workflow pause and involve a person?
Fallback path: What should happen when the workflow cannot proceed safely?
Clear pass conditions

Representative real-work test cases

Case	Input or condition	Expected behavior	Reviewer
Routine / high frequency
Missing / ambiguous information
High consequence / sensitive
Outside approved scope

Operationalize: Make the tested workflow usable and sustainable

Operationalize a tested workflow by packaging it so people other than its creator can use it, placing it where work already happens, supporting intended users, assigning access and ongoing ownership, and measuring evidence that guides what happens next. An announcement or one-time demo is not operationalization.

Decision	Notes
Intended users, approved scope, limits, and placement in existing work
Workflow owner + access: Who is accountable for the result, ongoing follow-through, and access?
Approvers, technical or governance partners, and support roles
Repeatable steps, examples, guidance, and support for first use
Support + feedback route for questions, overrides, failures, and suggestions
Evidence: value, reliability, control, and readiness
Review date and decision owner
Evidence-based recommendation: Stop, Revise, or Expand

Operationalize: Questions to consider

Can intended users repeat the workflow safely without relying on its creator?

Is the intended team result improving, not just product use?

Where do reviewers override or repair the output?

Are support needs, risks, or unintended effects increasing?

Who has authority to revise, expand, pause, or retire the workflow?

Choose your next unfinished decision

Decisions	Notes
Starting stage and first scope: Design / Develop / Operationalize
Desired outcome
AI and human boundary or other unresolved decision
First useful version or next requirement
Stop, ask, or escalate condition
Workflow owner
Evidence: value, reliability, control, and readiness

Next action

Share the relevant notes with the person who can validate the decision or take the next action. Depending on the unfinished step, that may be the workflow owner, an intended user, a business approver, a technical or governance partner, or a Transformation Leader. Record what changes and what must happen next.

Toolkit handoff

These resources are optional, adaptable starting points. Use only the resource that matches your next unfinished decision, and use your organization’s approved brief, ticket, checklist, standard process, or other format when it serves the same purpose.

Design: Choose the workflow — Workflow Starter Worksheet . Identify a recurring, multi-step workflow with meaningful team value and a manageable first scope. Capture what happens today, what should improve, and who understands or depends on the work.

Design: AI Workflow Design Coach . Map the current process, define the intended outcome and boundaries, clarify what AI may support versus what people must retain, and identify the users, owners, approvers, and partners involved.

Develop: Build and test it — PRD + Test Case Generator . Turn the design into concrete requirements and a first useful solution. Define approved tools, information, access, required and prohibited behavior, review and escalation, then test routine, variable, ambiguous, and high-risk cases.

Operationalize: Prepare it for reuse — Workflow Packager . Package the intended users, approved scope and limits, sources, repeatable steps, human-review gates, owner, access, supporting assets, help route, and change process so someone other than the creator can use it.

Operationalize: Workflow Adoption Plan . Place the workflow where the work already happens, support intended users through first use, define ownership and maintenance, establish feedback and escalation routes, and set the measures and review point that will guide what happens next.

Operationalize: Gather appropriate evidence of value . Use the Evidence Menu to identify useful signals across adoption, efficiency, quality, safe operation, and team outcome. Use the Evidence Log to record actual observations, measures, sources, exceptions, and lessons over time.

Operationalize: Activation Challenge template + Workflow Evidence Coach . Turn the available evidence into a clear story, distinguish supported claims from estimates or hypotheses, identify what to capture next, or prepare an internal case to continue learning, run a pilot, expand to a defined group, or consider broader scale.

Keep applying the method

Earn the Activator Labs Foundations badge: Attend at least 60% of the live Activator Labs 101 session.

Keep building: Use this workbook or the matching resource in the Champion Community on OpenAI Academy.

Share what works: Submit an applied workflow through the Monthly Activation Challenge. A qualifying submission shows the real team or functional workflow, what changed or was built, responsible AI and human boundaries, and credible evidence of value or improvement.

Earn the Agent Activator badge: Qualifying submissions are reviewed before the badge is awarded. A video is optional; a clear written submission can provide the same evidence.