Exploring AI Agents - Toward Autonomous Subject

In this lesson, we will explore the concept of agents, a pivotal advancement in AI systems characterized by their autonomy in solving assigned tasks. As agents are at the cutting edge of technological research, the definitions and best practices surrounding them are continually evolving. Let’s explore some reliable patterns to deepen our understanding of how these agents operate.

Agent are at the technical frontier, definitions and best practices are not fully settled.

Here some resources to get you started on the subject:

AI agents are equipped with various tools to enhance their autonomy and functionality. These tools can be used dynamically as the situation demands. For instance, in our demonstration application, which features a chat functionality, the agent can utilize external APIs or simple Python functions to perform tasks.

---
title: Exploring AI Agents - Toward Autonomous Systems
isPublicLesson: true
useMdx: true
privateVideoUrl: https://stream.mux.com/ph5pOMede5jHNHu6MgY9dSGTccMDyH18fb5Gdco1qrk.m3u8
---

# Exploring AI Agents - Toward Autonomous Subject

In this lesson, we will explore the concept of agents, a pivotal advancement in AI systems characterized by their autonomy in solving assigned tasks. As agents are at the cutting edge of technological research, the definitions and best practices surrounding them are continually evolving. Let’s explore some reliable patterns to deepen our understanding of how these agents operate.

#### App Preview

You can preview the demo app on Replit:

<ReplitEmbed src="//replit.com/@louissanna/module-7"/>

:::info Agent - a fast moving field

Agent are at the technical frontier, definitions and best practices are not fully settled.

Here some resources to get you started on the subject:

-   [Andrew Ng: the dawning age of agents](https://www.deeplearning.ai/the-batch/the-dawning-age-of-agents/)
-   [Nvidia: introduction to LLM agents](https://developer.nvidia.com/blog/introduction-to-llm-agents/)
-   [Andrej Karpathy: LLM OS](https://twitter.com/karpathy/status/1723140519554105733)

:::

### Tool Usage Pattern

AI agents are equipped with various tools to enhance their autonomy and functionality. These tools can be used dynamically as the situation demands. For instance, in our demonstration application, which features a chat functionality, the agent can utilize external APIs or simple Python functions to perform tasks.

Let’s consider a practical scenario. Suppose we ask the agent, "What's the average temperature of Tokyo and Paris?" Current large language models (LLMs) do not have access to real-time data and struggle with complex mathematical operations, which are essential for answering such a query. However, our agent can dynamically call an external API to fetch the temperature in Tokyo and another for Paris, and then calculate the average.

Louis Sanna

Exploring AI Agents - Toward Autonomous Systems

Welcome back. In this lesson we will discover agents. Agents are a new type of

where we give autonomy to a model to solve a given task. Agents are very much

frontier. Under every research and definition and respect list are not fully so

at a few relabel patterns to gain a better understanding. The two patterns I

are the tool usage and the self correction sometimes called refriches. Tool

Tool usage consists to give to the model several tools to use as needed. For

can see our demo app. We have a chat and this chat may use an external API some

function. Now let's say we ask a question. What's the average of a temperature

That cannot be responded by current LLM as LLMs do not have real time and are

mathematics. So let's take as our model. What does it do? It uses tools on a

So here we have a first tool called to an external URL for the Tompy Rachael in

same in Paris, an addition and a division. And I could ask for the weathering

and I will get the renewed action. So that's the tool usage pattern that works

is the self correction reflection pattern. Sometimes when you give a task to

it will fail on the first time. Let me show you. Here we ask for some code to

it filed with similar G-seconds and it executed and it failed because the

At what we did we gave it to the model initial task, how it failed and the

It made a certain implementation which is this time takes advantage of dynamic

be faster. That's a very powerful pattern that can increase the matrix of your

lot. Let me mention, briefly some exploratory pattern that do not work very

but that made the future. The first is planning when you ask the model to plan

future action and then to execute and the second is multi-algian collaboration.

Well let's say we will build a coding engine, a product manager engine and then

collaborate. Current generation models struggle a bit but in the future we

pattern to work. And last but not least I wanted to show you a visionary future

Andres Kerpatih famous machine learning researcher describing a vision of a

model are at the center of a new stack software 2.0 where you have ALM which is

administrator and then you have all the other services, you have memory, you

execution in your classical computer, you have specialized model, video model,

and you have some external access, kson api, browser and so on. Thanks a lot

we will start building our tool usage pattern. See you soon!

Learn

The newline Guide to Building Your First GraphQL Server with Node and TypeScript

Teach

Amelia Wattenberger

Author of Fullstack D3

Community

Masterclasses

Tutorials

Fullstack React with TypeScript