Replit Agent
Replit's commercial terms say private customer content and outputs are not used to train or improve AI features, while publicly published content may be used, and AI-integration endpoint privacy differs by plan.
Private customer content is excluded from training. Publicly published apps and content may be used to improve the service, including LLM training. Paid AI-integration endpoints have training disabled by default; free endpoints may allow provider training and publication.
Keep projects private rather than publishing publicly. Prefer paid or Enterprise zero-retention integration endpoints rather than free endpoints.
Customer content is private by default under the commercial agreement. Agent checkpoints can capture full project state, including files, conversation history, connected databases, and agent memory.
The reviewed docs do not quantify Agent checkpoint retention. Checkpoints are stored to support rollback. Enterprise AI-integration endpoints are described as zero-data-retention.
The reviewed commercial and integration docs do not provide one general human-review statement.
Foundational reference pages for teams adopting, deploying, and operating production AI with inspectable controls.