What is an agent?
That’s what Day 3 of Kaggle’s Gen AI Challenge is all about.
An agent is a system that observes its environment, plans actions, uses tools like APIs, functions, or data stores, and acts autonomously to achieve a goal – often over multiple steps (see whitepaper from Google below).
The cognitive architecture of an agent consists of three essential components:
a model (like a language model),
tools (like APIs or functions), and
an orchestration layer that coordinates reasoning and action.
You can build such agents using tools like LangChain and LangGraph.
The full whitepaper from course day 3: https://www.kaggle.com/whitepaper-agents