Inference

The process of running a model to produce a prediction or decision — the compute step behind every agent action.

Inference is the compute step of running a trained model to produce an output — the prediction, classification, or generated text that drives a decision. Every tick, an ForgeAI agent performs inference over the current game state and its playbook to produce its next action. Inference latency and cost directly constrain how quickly and how often an agent can act.