← All agents
AI agents · Product team · Feature Launch Readiness Agent

Do the whole job with me

Same job, two agents: one asks after every step, one asks once

In short

Two AI helpers plan the same new button.

One asks you after every step.

The other only asks before it saves.

Look at both of them below.

Think of two people helping you. A new intern checks with you after every task: "Done with this — shall I carry on?" An experienced colleague does the routine parts alone, but still asks before sending anything out for you.

An agent is an AI that does several steps in a row. Both agents here do the same 3 skill jobs: find what's missing, write a first draft (a short launch plan, saved as a file) and run the 6 launch checks. The only difference is how often they stop to ask you.

One job, two ways

Controlled vs autonomous Both agents do the same 3 steps. The controlled agent stops to ask you after step 1, before saving the file in step 2, and after step 3: 3 stops. The autonomous agent only stops once, before saving the file: 1 stop. Controlled Autonomous 1 Find what's missing Stop: go on? 2 Write the plan Stop: OK to save? 3 Run the 6 checks Stop: all done? 3 stops 1 Find what's missing 2 Write the plan Stop: OK to save? 3 Run the 6 checks done 1 stop

Controlled stops after every step. It even refuses "just do it all at once" — the stops are what this version is for.

Autonomous doesn't stop between routine steps, but it still asks before it saves the file. Fewer stops doesn't mean fewer rules.

The lesson: "autonomous" is not a yes/no switch. It's a choice about where a person checks the work — and saving or sending something is always one of those places.

What we gave both agents

The same one-line request as before — this time for the Orders page, with nothing else to go on.

To the controlled agent:

Use the Feature Launch Readiness Agent (controlled) on this: "Add CSV export to the Orders page, launching next Tuesday." Draft the brief to `launch-brief.md`.

To the autonomous agent:

Use the Feature Launch Readiness Agent (autonomous) on this: "Add CSV export to the Orders page, launching next Tuesday." Draft the brief to `launch-brief.md`.

What both of them knew:

Ask: "Add CSV export to the Orders page, launching next Tuesday."
No PRD, no rollout plan, no success metric mentioned.

A "CSV" is a simple file that opens in a spreadsheet. A "brief" and a "PRD" are both kinds of written plan. A "rollout plan" says who gets the new feature first. A "success metric" is a number that shows it worked.

What the real agents did

Two real runs, one per agent. Each time an agent stopped to ask, our reply is shown too.

Controlled

    Loading the recorded run…

    Autonomous

      Loading the recorded run…

      Worth noticing: the controlled agent's own instructions say the 6 launch checks should come in a reply of their own, not in the same reply that saves the file. In this run it saved the file and ran the checks together. It still stopped 3 times, but it didn't follow its instructions exactly. Even a good agent is worth reading with care.

      This page never talks to an AI. It shows runs that were recorded earlier, exactly as the agents wrote them — except two personal details we took out: a folder name on our computer, and a time zone.

      See the controlled agent's recipe card on GitHub See the autonomous agent's recipe card on GitHub Use them in your own AI tool