Inference
Also known as: Prediction, Model execution
When a trained AI model makes predictions or generates outputs based on new input. Inference happens every time you use AI.
Context
Inference happens every time you use AI. When your email tool suggests a subject line—that's inference. When a chatbot answers a customer question—inference. When your CRM predicts lead scores—inference. Training happens once (or occasionally), but inference happens millions of times. That's why AI services often charge separately for training vs. inference.
Examples
- 1ChatGPT generating an email response (inference from its training)
- 2Image recognition identifying your logo (inference on new photos)
- 3Predictive lead scoring calculating conversion likelihood (inference on new data)