Refinement Loop & Trigger Routing

The Refinement Loop is how CoffeeBreakAI learns from feedback and improves results.

Review-Driven Refinement

When a reviewer rejects or requests changes:

• Their comments are captured as structured feedback.
• A new task is created with Trigger Source: ReviewComment.
• A refinement agent uses this feedback to adjust the output.

Iterative Cycles

Refinement can happen in multiple cycles:

1. Agent produces an initial result.
2. Reviewer provides comments.
3. Agent refines the result using comments and capability hints.
4. Reviewer verifies and approves.

Benefits

• Reduces the back-and-forth manually copying feedback into prompts.
• Builds a history of decisions that can inform future tasks.
• Keeps humans in control of quality while still benefiting from automation.

The refinement loop is central to how CoffeeBreakAI turns feedback into better outcomes over time.