Create your own design team using agents

Do the work, step by step

1. Begin with three roles: Research Analyst, Interaction Designer and QA Critic. You can run them sequentially in separate conversations; autonomous multi-agent execution requires an environment that supports orchestration.
2. Give every role the same approved brief plus a narrow contract: inputs, output format, allowed tools, stop conditions and reviewer. The analyst produces an evidence ledger; the designer consumes the approved ledger and produces a flow; the critic checks it against the brief.
3. Add a coordinator that tracks version IDs, dependencies and unresolved questions. Stop if input is missing, roles disagree on a product rule, or the time or cost budget is reached. Limit revisions to two rounds before human review.
4. Approve each handoff yourself. Only then add UX Writer or Accessibility Reviewer roles. Prevent multiple agents from editing the same artifact at the same time.

A prompt to adapt

Replace the bracketed inputs with your own context. Copy this prompt into your chosen tool.

You are the Interaction Designer. Input: approved brief v1 and evidence ledger v1. Output: a booking flow with states, assumptions and source IDs. Do not create research findings or change product rules. Stop and ask the human lead when requirements conflict. Handoff to QA Critic with a version ID, changed decisions, unresolved risks and acceptance checklist. Maximum two revision rounds; no external publishing.

Your practice assignment

Run the three roles on the booking task. Intentionally omit a cancellation rule and check that the designer escalates rather than inventing one.

Source and teaching context

Anthropic: Cowork and chat are now one Claude. Product and method references checked October 24, 2026. Exercises are UX Design Coach teaching recommendations, not measured tool benchmarks. Interfaces and plan access can change.