:IMAGE

:INFO [valign:v_center]
Agent, take the wheel.
Agents are here to manage generic workflows from beginning to end. The more specialised the workflow is, the more expertise the agent can get.

:INFO [valign:v_center]
The 3 Core Parts of an AI Agent
1. Brain (LLM)
2. Tools (APIs, browser, database)
3. Memory (context over time)

:IMAGE

:CHECKLIST
How it works
[ ] You give a goal
[ ] It breaks goal into steps
[ ] It chooses tools
[ ] It checks results
[ ] It repeats until done

:NOTE.half
That loop = autonomy.

:INFO [valign:v_center]
Real World Example
- Searches web
- Extracts data
- Structures comparison
- Outputs summary table
- ...

:LINK.half
https://en.wikipedia.org/wiki/AutoGPT
AutoGPT (GitHub – autonomous AI agent)

:LINK.half
https://github.com/martimfasantos/ai-agent-frameworks
martimfasantos/ai-agents-frameworks

:LINK
https://github.com/HeyNina101/real-world-llm-agents
HeyNina101

:LINK
https://www.youtube.com/watch?v=fXizBc03D7E

:INFO [valign:v_center]
What's the Catch?
Agents aren't perfect, their autonomy usually means you will have to double check the results.

- Can make wrong decisions
- Can misuse tools
- Expensive (multiple model calls)
- Hard to debug

:INFO [valign:v_center]
Why Everyone Talks About Agents
We are shifting from having a “Software you use”
 to having a “Software that works for you.”

Having the speed of a machine and the brain of a human would allow workflows to be exponentially faster.

:IMAGE

:INFO [valign:v_center]
What Should You Learn Next?
Beginner path:

1. Prompting basics
2. APIs
3. Simple automation flows
4. Agent frameworks

Start small. Automate one workflow.