AI Assistant

The AI Assistant page uses a Bring Your Own API Key (BYOAK) model — you connect an API key from your preferred AI provider, and the platform uses it to power AI-assisted features like insights and chat. Your key is encrypted at rest and never exposed after saving.

Note

Your API key is encrypted and stored securely. It is never shown in plain text after the initial save. Anthropic recommends setting a billing limit on the key from your provider’s dashboard to protect against unexpected charges.

---

Status badge

Badge	Meaning
Not configured	No AI provider is set up — AI features are disabled
Not verified	Credentials were saved but the connection test has not passed
Verified	Provider is connected and credentials have been tested successfully

---

Supported providers

Provider	Example models	Extra field
OpenAI	gpt-4o, gpt-4o-mini, o1	—
Anthropic (Claude)	claude-3-5-sonnet-20241022, claude-3-haiku-20240307	—
OpenAI-compatible	Azure OpenAI, OpenRouter, Ollama, and any other OpenAI-compatible API	Base URL required

---

How to configure the AI Assistant

1. Select a provider

   Choose from the Provider dropdown. The model hint text and documentation link update to match the selected provider.

2. Enter the model name

   Type the exact model identifier in the Model field (e.g. gpt-4o). A Browse available models link opens the provider’s model list in a new tab.

3. Enter the Base URL (OpenAI-compatible only)

   If you selected OpenAI-compatible, a Base URL field appears. Enter the API endpoint for your service (e.g. https://openrouter.ai/api/v1 for OpenRouter, or http://localhost:11434/v1 for Ollama).

4. Enter the API key

   Paste the API key from your provider into the API Key field. Click the eye icon to toggle visibility.

5. Set a monthly token limit (optional)

   Enter a number in the Monthly token limit field (e.g. 500000). Once this number of tokens has been consumed in a calendar month, AI chat will be blocked until the 1st of the next month. Leave blank for no limit.

6. Test the connection

   Click Test Connection. The system sends a live request to the AI provider using your credentials and model.

   • A green banner confirms the credentials and model are working.
   • A red banner shows the error from the provider (e.g. invalid key, model not found).

7. Save

   Click Save (or Update if editing). The status badge changes to Verified.

Tip

Test the connection before saving. An untested configuration may result in AI features silently failing for all users.

---

Token usage counter

When a configuration is active, a counter beneath the status badge shows Tokens used this month and, if you set one, the monthly limit. Check this periodically to stay within your expected usage.

---

Removing the AI configuration

Click Remove Config at the bottom-left of the page and confirm. All AI-powered features are disabled immediately until a new configuration is saved.

---

Who can change AI settings?

Only users with the Settings — edit permission can configure or remove the AI provider.