Operator status
AI settings
AI runtime usage can be toggled here. Provider, model, and API key remain server-side environment configuration.
Disabled
Runtime status
Not usableSafe configuration snapshot. GEMINI_API_KEY remains server-side only and is never shown in the browser.
AI enabled
no
Provider
gemini
Model
gemini-flash-latest
Mode
disabled
Max rows per upload
200
API key configured
no
Usable
no
AI is disabled.
AI can run for all companies or uncertain rows only depending on runtime mode, remains capped per upload/provider/model/prompt/mode, and is displayed as a second opinion. The companies table and CSV export continue to use local score results plus SDR feedback overlays.
Use the controls below to enable or disable runtime AI usage and choose the assessment mode. Provider, model, and raw API keys are still configured in the server environment and are never rendered in this UI.
Company drawer actions queue AI assessment jobs only. A separate server-side worker must be running to complete them. Gemini quota/rate limits can pause jobs for retry; local scoring and SDR review remain official while jobs wait.
Queue worker commands
npm run ai:workernpm run ai:worker -- --uploadJobId=<upload-id>The worker reads server-side environment values, including the protected job secret. Do not put AI job secrets or provider keys in browser-visible variables.
Runtime controls
Toggle AI usage for uploads without exposing provider secrets. These settings override AI_ENABLED, AI_SCORING_MODE, and AI_MAX_ROWS_PER_UPLOAD at runtime.
Provider, model, and API key still come from server environment variables. The key is never shown or editable here.
Connection test
Sends one tiny generic prompt through the configured provider when AI is enabled and usable. It does not persist data or score companies.