Building Your Own AI Agent: A Guided Walkthrough

As part of their April session, New Jersey Code & Coffee presents Building Your Own AI Agent, a guided walkthrough of the core principles of AI agents: completions, tools, and the agentic loop.

Whether you're AI-curious or already shipping with LLMs, you'll leave with a solid mental model of how agents actually work under the hood.

We'll use the Anthropic Claude API with the TypeScript SDK as our foundation, building up from a simple completion all the way to a fully autonomous agent.

Bring your laptop if you want to follow along; there'll be one hands-on moment where we build the agent loop together.

All skill levels are welcome.

What You'll Learn:

• How LLM completions actually work under the hood
• How to give agents tools and let them decide when to use them
• The agentic loop — how agents reason, act, and iterate autonomously
• How to build a working agent from scratch using the Anthropic TypeScript SDK
• Mental models for thinking about agents that you can apply to any framework

What You’ll Need To Follow Along:

• A laptop with the latest version of Node.js and git installed
• An Anthropic API key with a few dollars of credits loaded

Doors open at 2pm, talk starts at 3, arrive early for networking and light refreshments.
Bring your laptop.

Speaker Bio:

Aza Walker is a self-taught software engineer with a fascination for computers and the intricate systems that make them work.
He enjoys designing and building large-scale applications, and currently works at ShipBob as a Senior Software Engineer, where he focuses on developing scalable microservices.