AI coding education and developer tools
AI Hero
AI Hero teaches developers how to work with coding agents, from clarifying requirements to implementing and reviewing code. Its offerings include a free email course, the self-paced AI Coding Crash Course, and AI Coding for Real Engineers, a two-week cohort combining asynchronous lessons with live office hours. Its mattpocock/skills library supplies reusable agent instructions for planning, specifications, task breakdown, test-first implementation and code review, with support for Claude Code, Cursor, Codex and Copilot.
Creator Matt Pocock teaches AI coding full-time at AI Hero. He previously created Total TypeScript, worked as a developer advocate at Vercel and served on the XState core team. His skills encode engineering practices in plain files: developers can install and edit them within a project. The related Sandcastle TypeScript library orchestrates coding agents in isolated environments, manages branches and merges their commits back. It supports parallel agent work and review pipelines, with built-in Docker, Podman and Vercel sandbox providers.
As of August 2026, AI Hero reported 8,500+ engineers trained in cohorts. Its distribution combines 25 free skills with commercial training: the AI Coding Crash Course contains nearly 60 lessons across six sections and offers team licensing.
3 talks
Newest first
"Software Fundamentals Matter More Than Ever" — Matt Pocock
◆◆◆Matt Pocock
AI Engineer Europe 2026 · 18:26
Full Walkthrough: Workflow for AI Coding — Matt Pocock
◆◆◆Matt Pocock
AI Engineer Europe 2026 · 1:36:30
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Start here
- "Software Fundamentals Matter More Than Ever" — Matt Pocock
Start with the essay based on Matt Pocock's talk to understand how shared domain language and frequent feedback help prevent generated code from drifting away from the intended system.
Matt PocockAI Engineer Europe 2026
- Full Walkthrough: Workflow for AI Coding — Matt Pocock
Read the essay based on Matt Pocock's walkthrough to see how a Kanban issue graph organizes unattended work and why he recommends observing single iterations before trusting a longer agent loop.
Matt PocockAI Engineer Europe 2026
- Building Great Agent Skills: The Missing Manual
Use the essay based on Matt Pocock's talk to understand the context cost of model-invoked skills and how to decide which instructions belong in the main procedure or conditional references.
Matt PocockAI Engineer World's Fair 2026
Messages from the stage
Turn requirements into executable work
Pocock's workflow moves from stakeholder interviews to a product requirements document and dependency-aware vertical slices. Autonomous implementation sits within a cycle of testing, isolated review, and human QA.
Make skills focused and selective
Skill design involves choosing who invokes a skill, separating procedural steps from supporting references, and removing duplicated guidance. Pocock also describes compact steering language and sequential skills for directing agent behavior.
Design interfaces that support testing
Pocock connects deep modules and simple interfaces to meaningful tests at module boundaries. He places interface design under substantial human control while allowing agents more responsibility for internal implementation.
Affiliations reflect each recorded session, not necessarily current employment.