Lyra best practices, glossary, and FAQ

This page covers best practices for Lyra admins, a glossary of key terms, and frequently asked questions.

On this page: Best practices | Glossary | FAQ

Best practices for Lyra admins

Start small, iterate often. Launch with one or two coaches focused on your highest-value use cases. Monitor usage, gather feedback, and refine before scaling to more coaches.

Curate your knowledge base. Quality over quantity. Remove outdated content regularly. A coach that gives a confident but wrong answer (because its knowledge base is stale) is worse than one that says "I don't have information on that."
Write purpose statements like job descriptions. The more specific your purpose, the better the coach performs. Include what the coach should do, how it should behave, what tone to use, and when to escalate or defer.
Test from the user's perspective. Before launching a coach, test it with the kinds of questions your users will actually ask. Check whether the responses are grounded in knowledge base content or falling back to generic AI responses.
Monitor and iterate. Review coach analytics (messages sent, threads created, active users) to understand engagement. Low engagement might mean users don't know the coach exists, don't find it useful, or the coach might not be relevant enough for learners.

Glossary

Term	Definition
Coach	An individual AI agent configured with a specific persona, purpose, and knowledge base. Each coach serves a defined use case.
Knowledge base	The collection of documents and content sources that a coach draws on to answer questions.
Purpose	The instruction that defines a coach's role, behaviour, tone, and scope.
MCP (Model Context Protocol)	The technical protocol that enables Lyra to connect to external data sources and tools.
Thread	A single conversation between a user and a coach.
Session	A user's interaction period with Lyra, which may include multiple threads across different coaches.
Widget	A visual, interactive UI element (card, button, tile) that can appear within the chat experience.

Frequently asked questions

Can users talk to multiple coaches?

Yes. Users can switch between coaches depending on what they need help with. Each coach maintains its own conversation history.

Can a coach access content from outside Fuse?

Not at the moment. We are actively working on external content integrations for Lyra to access in the future.

What happens if a coach can't find an answer in its knowledge base?

If the knowledge base doesn't contain relevant information, the coach will either let the user know it doesn't have specific information on that topic, or fall back to general knowledge (depending on how you have configured the coach). The behaviour depends on how the Purpose is configured — you can instruct the coach to always ground its responses in the knowledge base and clearly flag when it's unable to find relevant content.

Can we control which users see which coaches?

Yes. Coach visibility can be managed through community-based access controls within Fuse.

Does Lyra remember previous conversations?

No. Lyra cannot remember previous conversations at the moment. This will be a future roadmap feature.

What languages does Lyra support?

Lyra currently supports English, Spanish, French, German, Italian, and Portuguese. The interface adapts to the user's browser language, defaulting to English for unsupported languages.