Switch to Your Own LLM Provider
FunnelStory can run its AI on your own LLM provider instead of the platform default — useful when you have a negotiated contract, a data-residency requirement, or a custom endpoint. This guide registers a provider and points FunnelStory's model slots at it. The full reference is AI Providers.
What you can connect
- OpenAI, Google Gemini, or Google Vertex AI. Vertex also needs a project and region/location.
- A custom OpenAI-compatible endpoint, by setting an
api_baseon an OpenAI provider.
1. Add a provider
Go to Admin → AI Providers → Add Provider, choose the vendor, and enter its credentials (for Vertex, the project and region too).
2. Map the model slots
FunnelStory uses three slots — Large, Small, and Embedding. Point each at a model your provider offers. The Large slot does the heavy reasoning (agents, investigations); Small handles lighter calls; Embedding powers semantic search.
3. Custom endpoint: re-enter the key when you change the base URL
If you use a custom OpenAI api_base, note that changing the api_base requires re-entering the api_key — the two are validated together. Plan for it when you rotate endpoints.
4. Verify
After saving, confirm FunnelStory is using your provider (for example, run an agent or an AI summary), and check the Audit log for the provider change.
Related
- AI Providers — the full reference for vendors and model slots.
- Audit log — a record of configuration changes.