Be part of JetBrains PHPverse 2026 on June 9 – a free online event bringing PHP devs worldwide together.

Browse all series

So You Want to Build an Agent

You're likely using an AI agent every day at this point, but do you truly understand how it works under the hood? What separates an agent from a general chat? What's a tool? How do we provide the loop for executing these tools? In this series, we'll demystify some of these concepts by building a basic agent that is capable of interacting with your file system and triggering tools.

Progress

Series Info

Episodes
9
Run Time
1h 56m
Difficulty
Intermediate
Last Updated
Apr 16, 2026
Version
Latest
Play Series

Series Episodes

  1. The Essentials (7)

    1. Your First AI Response

      Before we discuss agents, let's ensure that everyone is on the same page. We'll begin by building a simple command that sends a prompt to AI and prints its response.

    2. Have A Dialog

      In the previous episode, we captured the AI response...and that was the end of the conversation. To truly have a dialog, we should maintain and track the history.

    3. The Agent Loop

      A key component to an agent is the loop that drives it. I like to think of this loop as the agent's inner dialog. Let's discuss it!

    4. Extract A Tool Class

      The second piece of the agent puzzle is tooling. By default, of course an agent can't interact with your local filesystem. Instead, we must provide that functionality as a tool that the agent can be then request.

    5. Make a Revenue Tool

      A few episodes back we asked the AI about last quarter’s revenue, and, of course, it had no clue. Let's fix that by providing a Revenue tool that it can trigger.

    6. Structured Output

      Now that we have our basic agent loop working, we are going to tighten things up by introducing structured output so each agent can be forced to return the exact shape we expect.

    7. Filesystem Tools

      Let's review some of the core filesystem tools that any agent will require. This includes everything from executing Bash scripts, to reading files, to performing searches.
  2. Going Deeper (2)

    1. Compacting

      Welcome to section two! Now that you know the essentials, let's dig deeper. First up, we need to structure our conversation history in such as way so as to grow endlessly and wreck our context window.

    2. Load a Guidelines File

      Next, we should teach our agent how to pick up on project-level guidelines (think CLAUDE.md or AGENTS.md) and fold them into its instructions.

Continue Learning

The Batteries-Included AI Toolkit thumbnail

The Batteries-Included AI Toolkit

With Jeremy McPeak

The Laravel AI SDK is a first-party toolkit for adding AI features to your applications. Using a helpdesk app as our playground, you’ll learn structured output, chat with memory, streaming, tool use, retrieval, and everything else you need to build AI features that are reliable, testable, and ready for production. This is an incredibly powerful toolkit, and I can't wait to show you what you can build. Let's get going!

  • 13 Episodes
Watch Me AI the UI thumbnail

Watch Me AI the UI

With Jeffrey Way

In this workshop, come along as I upgrade the visual design of our Laracasts team dashboard. As you'll see, design work can sometimes be a bit tricky to orchestrate. Can AI properly evaluate a screenshot and convert it to CSS? Well, yes...and no! But luckily, we have a variety of techniques we can use to increase our odds of success. Let's go!

  • 5 Episodes
AI Developer Workflows thumbnail

AI Developer Workflows

With Jeffrey Way

It's a brand new world. Suddenly, we find ourselves less writing code line by line, and more writing plans, coordinating agents, and building workflows. There's a lot to learn. But we gotcha!In this series, each episode will demonstrate one potential AI workflow that you might adopt in your own projects. Check back often, as we'll be adding new episodes every month.

  • 3 Episodes
Leveraging AI for Laravel Development thumbnail

Leveraging AI for Laravel Development

With Jeffrey Way

The web development world forever changed in December, 2025. This was the month when we realized just how powerful and useful LLMs had truly become. And there's no turning back. It's time to get on board! This train isn't stopping. Join me, as I get you up to speed in record time. Learn about Jetbrains AI, agents, prompting, vibe coding, developer-driven AI, Ralph Wiggum, and a wealth of tips and techniques to drastically speed up your workflow. Let's go!

  • 18 Episodes
Vibe Coding Workshop thumbnail

Vibe Coding Workshop

With Josh Johnson

Hi, I'm Josh, the creator of Wender. In this series, I'll teach you everything I've learned about how to treat vibe coding like a craft... rather than a magic trick. We're not chasing weekend Airbnb clones. No, we're interested in practical workflows for building better applications using AI. You'll learn about writing better prompts, what to do when things go sideways, and how to review diffs like a pro!

  • 11 Episodes