What happens when you reject an AI agent's work in Bees

Description and use

In Bees, rejecting an AI agent’s output triggers scoped feedback that updates only the current or future scheduled runs, keeping the base agent unchanged.

Problem

When an AI agent produces incorrect output, users need a way to reject the work and provide feedback that influences future runs without causing unintended changes.

Outcome

Rejection feedback is captured, scoped, and applied only to relevant runs, preventing unintended rule creation and keeping the base agent unchanged.

Workflow

• Define a goal and run the work agent
• Reviewer agent checks the result
• If rejected, reviewer sends feedback back to work agent
• User chooses "Reject" and provides a reason
• System asks if reason applies to this run only or future schedule runs
• If future runs, reason is added to the job’s playbook
• Specialist agent inherits base agent plus playbook notes
• Future runs use updated specialist agent
• User can view, edit, undo, or reset playbooks via Process Runs → Schedules

Evidence

If the result falls short, the reviewer sends the same item back to Work with specific feedback, and the next attempt sees both the old output and the notes.

Source

On a scheduled job, Bees asks where the reason should apply: This run only or Future schedule runs.

Source

That playbook belongs to a specialist. The first time an agent handles a named schedule, Bees creates a specialist for that pair.

Source

You can edit it, undo the last change, or reset it to empty and go back to the base agent.

Source