πŸ“

Intro to AI agents

image

My TL;DR about agents

πŸ”— Watch this https://youtu.be/oP6DS_x5K0Y?si=pTzcxmaBR2yIqebC

At the heart of agentic AI are autonomous agents that can:

  1. Perceive: Interpret input from the environment.
  2. Plan: Generate strategies to achieve goals.
  3. Act: Execute actions and interact with the world.
  4. Learn: Improve decisions based on feedback.

Basically, AI agent wraps LLM with a few components:

  1. Tools - to act, external capabilities to perform some action other than just generating some text (eg. web search, call APIs, run code in a sandbox, send email), mostly MCP tools. Tool = brain.
  2. Memory - to remember things, long term memory, because agent cannot remember everything in 1 context window (context window = short term memory)
  3. Reasoning and planning loop - to get the task done, turns LLM into agent with continuous cycle, agent breaks big task into small sub-tasks (simulating thinking system #2 from "Thinking fast and slow"). The loop of "reAct": think β†’ act β†’ observe β†’ repeat.
  4. Perception - the ability to see and hear environment by monitoring chat/database/website.

AGENT = LLM + Planning + Memory + Tools

Remember: the most important component is reasoning loop πŸ”₯ which is a continuous loop of thinking and acting.

Must read:

How to build own agents?

Learn how to build MCP and agents:

Deep dive:

Check more about Agentic AI in

🌱ML & Agentic AI. ROADMAP
SuperMade with Super