Kelreva

Verify AI claims against real sources with one click

Kelreva scrapes the web live and brings you the clearest source and verdict, all with one click. Integrates seamlessly with ChatGPT, Claude, and Gemini. What you get: - Supported, partially supported, not enough evidence, or refuted results - A short explanation for each checked claim - Source links when relevant evidence is available - Direct highlighting of checked passages - User-initiated checks for selected text or a completed response The workflow: 1. Select: Highlight any text or click the Kelreva icon below a response. 2. Separate: The answer is broken into specific, checkable claims. 3. Search: Public web evidence is retrieved and ranked for relevance and authority. 4. Compare: Each claim is marked against the evidence and linked back to its sources. Privacy and requirements: Kelreva does not automatically send every page or conversation. A check begins only after you choose Verify and have consented to processing. Kelreva then sends the associated question, the response text selected for checking, a random installation identifier, and account-session credentials to the configured Kelreva backend. The backend uses Tavily to retrieve public web evidence and OpenAI to identify and evaluate claims. A Kelreva account and an internet connection are required. Do not submit sensitive or confidential conversations.