AI
How Rian uses your stored knowledge.
Knowledge retrieval
When a customer sends a message, Rian searches your knowledge base — documents, indexed site pages, and articles — to find the most relevant content before generating a response.
Search
The Storage overview includes a search bar. Type a query (e.g. "Summarise refund policy") and click Search. Quick chip suggestions appear below the search bar based on your existing content.
Memory tiers
Rian retrieves knowledge across four tiers with different latency profiles:
| Tier | Description |
|---|---|
| Hot | Active session context; sub-millisecond |
| Working | Short-term working memory |
| Warm | Cached knowledge; fast retrieval |
| Cold | Full knowledge base search; highest latency |
You can inspect these tiers live in the Agent Sandbox > Memory tab.
Improve accuracy
If Rian is giving incorrect or incomplete answers:
- Check AI Agent > Insights to see which sources were used
- Update or add content in Storage to cover the gap
- Re-test in the Sandbox
AI article generation
Use Create with AI inside any category to draft a new article from a prompt. Review and edit before saving.