Documentation
2. Setting Up an AI Provider
- Created
- Jun 5, 2026
- Updated
- Jun 5, 2026
The assistant talks to a language model through a provider you choose. Configure your provider and key in the AI assistant tab of Chart Settings (or from the assistant panel), then pick a model.
Supported providers
| Provider | Key | Connection |
|---|---|---|
| OpenAI | OpenAI API key | Direct from the browser |
| Anthropic | Anthropic API key | Direct from the browser |
| Google AI Studio key | Direct from the browser | |
| AWS Bedrock | Bedrock API key | Through a server proxy |
| Google Vertex AI | Service-account JSON | Through a server proxy |
Each provider offers a list of models; the workspace selects a sensible default and you can switch models per provider. Provider docs links are shown next to each key field so you can create a key.
How keys are used
- OpenAI, Anthropic, and Google are called directly from your browser using the key you enter.
- AWS Bedrock and Google Vertex AI are routed through a server proxy (server-side routes under
/api/ai/...), because they require server-side credentials. Vertex expects a service-account JSON key rather than a plain string.
Security: where your key is stored
Your API key is stored in your browser's local storage on this device, and (for OpenAI/Anthropic/Google) is sent directly from your browser to the provider.
- Use a key scoped to the minimum permissions you need, and prefer keys you can rotate or revoke.
- Avoid entering keys on shared or untrusted machines; clearing site data removes stored keys.
- The key is not synced across devices and is not shared between browsers.
The assistant can also verify a key with a lightweight connection test so you can confirm it works before chatting.
Next steps
With a provider configured, start a conversation.
Next: Chatting With Your Chart