Model Randomness vs State Mutation

LLM output randomness comes from sampling and inference behavior. IAP does not change that.

State mutation comes from changes to tracked inputs such as prompts, local memory files, tools, policies, or key configuration.

In the current beta, IAP helps by certifying ordered state checkpoints for a given agent_id. It does not make repeated model outputs identical.

Back to landing page