One thing to try next
Start with one real task.
Choose one developer, one repository, and one agent. Agree on permissions and review the work together before expanding the pilot.
A small pilot, with clear ownership.
Read the full playbook The AI adoption field guide Your team / your next step
The AI adoption quiz
Find yournext step.
Start with how your team works today. Leave with a clearer picture of where you are—and one practical move forward.
Four active steps. Start with your workflow.
Sheet Q-001 Find your starting pointChecklist
A little reflection.A clearer direction.
Use the full checklist to assess your team, or take the framework with you.
Open the full checklist Download the white paper (PDF)Sheet Q-001 Question 1 of 14Everyday use
Is this true for your team today?
A person regularly uses a coding agent on real, bounded work.
Is this true for your team today?
Each task has an identifiable human owner and acceptance criteria.
Is this true for your team today?
The agent's changes are checked before they are accepted.
Is this true for your team today?
The team understands the agent's access and how to handle questions or blocked work.
Is this true for your team today?
Independent agent tasks routinely progress at the same time in separate working contexts.
Is this true for your team today?
Someone can identify each task's owner, status, and branch, and coordinate review and integration.
Is this true for your team today?
People and agents share the requirements, progress, and decisions needed to coordinate a workflow.
Is this true for your team today?
Agents can delegate bounded work and use the resulting reports without a person relaying every handoff.
Is this true for your team today?
Dependencies, completion criteria, and ownership of failed or blocked work are explicit.
Is this true for your team today?
The team has defined which actions need human approval and how decisions can be reviewed.
Is this true for your team today?
People contribute domain judgment and evaluate the evidence before accepting the outcome.
Is this true for your team today?
An assistant carries a standing responsibility across runs, using retained context to plan and coordinate follow-up work.
Is this true for your team today?
That responsibility has a human owner, resource limits, approval boundaries, and a regular review of outcomes and exceptions.
One last question
How many engineers are on your team?
This does not change your step. It helps us suggest what to do next.
Answer for today, not where you want to be.
Q-100Your team’s current stepResult
A starting point for the conversation. A next step you can put to work.
- You are here 00 Gated Get access
- You are here 01 Assisted Work with an agent
- You are here 02 Parallel Run agents in parallel
- You are here 03 Supervised autonomy Work together
- You are here 04 AI-native Own ongoing work
What this looks like
You give an agent a task, answer its questions, and check its work—even from your phone.
See Glueprint in actionOne thing to try next
Give every task room to work.
Try two independent tasks in separate worktrees. Keep each agent’s branch, progress, and requests in view as you review the results.
Independent work, with a place for every agent.
Read the full playbookWhat this looks like
Several agents work at once in separate worktrees, on your machine or remote hosts. You assign the tasks and pass results between them.
See Glueprint in actionOne thing to try next
Bring your team and agents together.
Choose a shared task. Bring teammates and assistants into a channel, delegate bounded work with clear handoffs, and review the results together.
Shared context, bounded delegation, and review.
Read the full playbookWhat this looks like
People and assistants share context in channels while agents coordinate and delegate bounded work. People shape requirements, resolve exceptions, and review results.
See Glueprint in actionOne thing to try next
Turn recurring work into a standing goal.
Pick an objective your team keeps re-prompting. Give a persistent assistant the context, memory, and operating boundaries to carry it forward.
Continuity, with people setting direction.
Read the full playbookWhat this looks like
Persistent assistants own ongoing responsibilities: they remember past work, plan next tasks, and coordinate other agents across days.
Glueprint supports ongoing responsibilities today through persistent assistants with memory, routines, channels, tool permissions, and token budgets.
See Glueprint in actionRefine ongoing ownership
Keep intent and accountability together.
Review your ongoing responsibilities: who sets direction, what agents remember, and which actions need human approval. Make those expectations explicit across your team.
Clear responsibility as autonomy grows.
Read the full playbookQ-200 Put your next step into practice
Your agents. Your team.A shared place to work.
Talk through the next step for your engineering organization with us.
Based on the Steps of AI Adoption. Your step is the highest one whose statements are all true for your team.
Prefer the full checklist?