Agentic Coding
in Practice
By Matthew Sinclair
A hands-on, five-day course on building real software with AI coding agents — from the shifting landscape and the human-agent conversation, through architecture, process, and the reckoning of agent-built code, to running a fleet of agents as a single human. Built from 18 months of real development: 12+ repositories, 3,456 commits, 155 sessions, and one honest failure.
New here? Start Here — a short, free orientation. Or skip ahead to the syllabus or Day 1.
The five days
Five consecutive days, each a theory session plus hands-on practice. The arc runs from orientation to synthesis:
- Day 1: The Landscape — Where are we? The agentic ecosystem, the autocomplete-to-autonomous spectrum, and the gap between what the tools do and what we do with them.
- Day 2: The Conversation — How do we communicate with agents? Context engineering over prompt engineering, and the delegation model.
- Day 3: The Architecture — How do we structure code for agents? CLAUDE.md as living architecture, the Highlander Rule, and skills as procedural memory.
- Day 4: The Process — How do we execute durably, and where does it break down? Steel threads, the session lifecycle, scope discipline — then the reckoning: recovery and the generation trap.
- Day 5: The Hypervisor — How does one human run a fleet? Multi-agent agentic coding (MAAC): coordinating a fleet of agents through a shared whiteboard — the one-human software house.
Each day's hub links its theory, exercises, reference card, and that day's Insights. See it all at a glance on the course map.
What you'll get
Attendee Pack
The course companion: the full five-day programme — theory, exercises, and a reference card per day — ready to follow along.
Complete Reference
Everything in the Attendee Pack plus the 31-Insight reference, the Enterprise guide, and a downloadable pack for every day.
Insights Reference
The 31 Agentic Coding Insights (ACIs): the field-tested rules, a la carte.