Build custom AI #1 — Support agent (Python / LangGraph)

Join for an interactive, small-group session on building custom AI agents that actually work.

In this practical workshop, you'll:

• learn the core building blocks of agentic AI
• compare modern frameworks for structured, reliable execution
• build and test your own agent end-to-end

Hands-on environment:
You'll work inside a framework-light Agentic Workbench — a local playground for building and testing AI workflows. We'll begin with a pre-built support-assistant agent to understand how it works, then you'll quickly move on to building or extending your own custom agentic system.

You'll cover:

• LLM gateways (e.g. OpenRouter)
• structured outputs with Pydantic
• prompt engineering best practices
• evaluation methods

Requirements:

• bring a laptop
• comfortable writing and running Python
• be prepared to use your own API tokens — usage during the workshop should stay inexpensive (typically <£1)

You'll leave having:

• run enterprise-grade agentic patterns
• built a working agentic system from scratch
• gained clarity on how to choose and deploy frameworks

About the facilitator
Dexter is an independent consultant and workflow architect. He has delivered and advised on high-accuracy, agentic workflows for growth-stage and enterprise clients in financial services, energy, telecommunications, and digital marketplaces — often unlocking stalled projects in under two weeks. Recently, he has focused on CLI coding tools such as Claude Code to improve how code, documents, and notes are maintained.