•Featured
What Is An AI agent?
An AI Agent is a system that completes a task by using a Large Language Model (LLM) to decide what steps to take and when the task is complete.
There are many (disputed) definitions of what an AI Agent is, but here's mine: An AI Agent is a system that completes a task by using a Large Language Model (LLM) to decide what steps to take and when it is complete.
To put it more technically, to be an agent, the system has to call an LLM, which has access to tools it can optionally call and runs in a loop until the LLM deems the task as complete.
This diagram from Harrison Chase (CEO of LangChain) clearly shows the boundaries between human-driven systems (steps 1 - 4) and agentic systems (steps 5 - 6) - using "cycles" as terminology here for loops.
