The Science One framework introduces a Chain-of-Evidence mechanism to make autonomous research verifiable. It forces agents to link every claim to a specific, traceable piece of evidence. This prevents hallucinations during complex scientific discovery. Practitioners can now audit the exact reasoning path used by Google AI agents to reach a conclusion.