Several Venvera features are powered by AI: the AI Policy Review, the AI Evidence Assistant, and the Virtual CISO chat assistant. They all share one setting: your organisation's AI provider and API key, configured once in Settings, AI Assistant. This article covers how to set that up.
Choosing a provider
Venvera supports three providers. Pick the one you have an account with:
| Provider | Key prefix | Default model |
|---|---|---|
| Claude (Anthropic) | sk-ant- | claude-sonnet-4-5 |
| OpenAI (ChatGPT) | sk- | gpt-4o |
| DeepSeek | sk- | deepseek-chat |
Adding your key
Go to Settings and find the AI Assistant section. Configuring it requires the settings.manage permission.
Select your provider and paste the API key. Venvera checks the key prefix, then live-validates it against the real provider before storing it encrypted.
If the key is valid it is saved encrypted; if not, the save is rejected. Venvera never displays the key again; the status view only shows whether a valid key is configured.
Shared frameworks
The shared frameworks setting controls which frameworks the assistant is allowed to read data for. The assistant only ever sees aggregate counts and scores for those frameworks, scoped to your organisation by the database's row-level security. It never reads individual records. This ties into Company Groups, so you can decide how much of your aggregate posture the assistant may use.