Original subtitles

- [Narrator] Now we will learn how we can design

one or more LLMs to work together

to become successful Agent AI.

There are many different ways we can design this.

This is called Agent design framework.

This is an evolving area of Agent AI research.

You may add your own design approach

as you begin building out agents

for your customer use cases.

I'm going to teach you the common agent design

frameworks that you will need.

Reflection is how the LLM ponders its own results

and improves it.

Tools, we've been saying Agent AI uses tools.

The tool use of an LLM could be a web search

to gather data and to process the information.

Planning is how the LLM comes with a step-by-step plan

to execute a task.

Multi-agent collaboration is when multiple

agents work together.

They could debate with each other, they could collaborate

to arrive at the best solution for a problem.

They might have individual goals,

or they might have one common goal for the Agent AI system.

Now we have learned multiple design approaches

to build Agent AI.