So, LLM inference is something where we use trained LLMs like GPT-6, GLM-5, DeekSeek-V4.1 to generate outputs or texts from user inputs or prompts. There are many examples like custom support chatbots - when the chatbots are reply or taking any actions based on user’s input or text, those actions are done through the process of inferencing. Same goes with ai agents but its just more steps and workflows in source