A client in onboarding asked to be able to export the AI conversation logs natively from the platform the actual message history from conversations handled by the agent, not just performance metrics. The goal is to review how the AI responded and spot opportunities to improve the playbook: questions answered poorly, conversations that could have been resolved without a handover, and prompts that need adjusting. Metrics show how the agent is performing; the logs show why. Reviewing conversation by conversation in the Inbox doesn't scale, so today the only alternative is the Platform API.