Intelligence, with human agency.
Design AI experiences that communicate limits and keep meaningful control with the person.
Interaction principles
- Identify AI-generated output and show the information used to produce it when useful.
- Make uncertainty visible. Avoid presenting generated claims as verified facts.
- Preview consequential actions and obtain task-appropriate confirmation.
- Provide stop, retry, edit and human escalation paths.
- Keep prompts, documents and personal information within approved service boundaries.
- Explain what is saved and let people review or remove their history where the product supports it.
Skunkie AI learning companion
For Academy journeys, introduce Skunkie as the AI learning companion. Explain how an assessment or learner input will shape recommendations. Give the learner a way to change preferences and inspect their roadmap.
Use an accessible stream of complete messages, clear loading feedback and a recoverable service-unavailable state. Keep learning options available when the companion cannot respond.
A reviewable action example
Show the chosen topics, intended outcomes and saved learner preferences. Let the learner edit or confirm the proposed roadmap.

