Skip to content

Run a Business Loop

A Business Loop gives your agent ongoing responsibility for progress toward a business goal. Each pass resumes current decisions and work, examines what changed and chooses the next useful action. You can run it in conversation or arrange a recurring check when the environment supports one.

Bring the current product, relevant research, active commitments and nearest desired outcome. Explain the horizon, budget and operating constraints that matter. Your agent should distinguish your goal from your diagnosis of the obstacle and recover earlier decisions before collecting more information.

Review our current business against the goal of helping customers complete and use their first report. Read the existing product notes and evidence. Recommend one primary focus, compare the nearest credible alternative, and prepare the next piece of work. Keep new spending, outreach, publication and product changes pending until their scope is agreed.

A useful result names the focus, why it matters now, what the agent can produce, what you need to decide or do, and when its effect can be assessed.

The next action might be a product improvement, acquisition test, operating repair, missing piece of evidence or continuation of work already underway. Waiting for an agreed evaluation point can also be appropriate.

For example, evidence that customers lose their work when returning could justify a continuity repair. Evidence that the journey works but the intended audience cannot find it could support an acquisition test. These are illustrative possibilities; the recommendation should follow your actual evidence.

Your agent compares credible alternatives through expected contribution, confidence, time, cost, risk and ability to execute. One primary focus can coexist with urgent customer obligations or useful parallel work.

Once agreed, keep the concrete result, owner, scope, limits and next checkpoint in the existing plan. The agent continues authorized work through the relevant capability. For founder-owned work, agree the needed result and follow-up point.

Observe the evidence that could change the next action. A campaign stop condition may need attention sooner than the overall business review. A product improvement may need time before customer effects become interpretable. Preserve the original experiment question and criteria as results arrive.

At the next pass, expect a concise update on meaningful changes, progress, observed effects and any decision needed. The agent should recover the last verified state and reconcile uncertain execution before repeating it.

Agree the schedule, destination, sources and action scope before enabling recurring work. The runner must remain available, with the project context accessible; local scheduling requires the relevant runtime to be awake and able to execute. Registration alone does not prove a scheduled run succeeded.

Use Continuity between sessions to establish that arrangement. A manual Business Loop remains useful whenever scheduling is unavailable or unnecessary.