$ learntobuild --start # the channel is back, as a full hub

Learn to build —
and actually ship — with AI.

The free hub for going from “the agent wrote 70% of it” to a real thing that’s live and getting used. Taught the way it’s actually done — by a Forward Deployed Engineer who ships enterprise AI agents for a living.

Start with Track 0 → See the project ladder

Claude Code Codex opencode Cursor MCP · evals · orchestration

Why this exists

Everyone can start a project with AI. Almost nobody ships one.

The tools scaffold ~70% of an app in minutes — then the tutorials leave you there. Auth, a real database, deploys, debugging when it breaks off the happy path. That last stretch is where projects die, and it’s exactly what this hub is built to teach.

Not “learn to prompt.” Not “learn to code” in the abstract. Learn to finish.

Three ways in

Start free. Level up when you want accountability. Bring me in when you want it shipped for you.

Free · Learn

The hub

Tracks, a project ladder with real “done” criteria, a curated resource library, and public build logs.

Browse the tracks →

Paid · Level up

Build-with-me

Live workshops + a community that carries you through the parts free content can’t: the finish, with accountability.

coming soon

Consulting · Work with me

Done for you

For teams that want it shipped this quarter. A productized audit → build sprint → retainer, run by an FDE.

See how it works →

What you’ll learn

Seven tracks, foundations → capstone. One transferable pattern, then each tool’s dialect — every project ending in a live URL or passing evals.

T0 Foundations What an AI coding agent actually is, the loop it runs, and the one mental model that makes every tool make sense. No building yet — just the ideas everything else stands on. live T1 Your first build Pick one tool and ship one real thing. You'll build the same starter project in Claude Code, Cursor, Codex, or opencode — so you feel the shared pattern instead of memorizing one tool. draft T2 Context & workflow The habits that make an agent go from frustrating to fast: a context file it reads, working from a spec instead of a wish, and building your own commands so your workflow repeats. T3 Real apps Where the tutorials abandon you and the 70% problem lives: databases, auth, secrets, real APIs, and deploying something that stays up. This is the track that teaches you to finish. T4 Agents, tools & MCP How agents get hands. Build your first MCP server, connect real tools to your editor, and make an agent that completes a task end-to-end instead of just answering. T5 Quality & evals How to know your AI is actually good — and stays good. Write evals, gate them in CI, review agent-written code, and keep cost and context under control. The senior skill. T6 Orchestration · capstone Put it all together: decompose a real product, run multiple agents against a spec in parallel, verify against acceptance criteria, and ship. Your proof-of-skill build.

You learn by shipping

A ladder of twelve builds, beginner → advanced. Each one is done when it’s real — live, tested, or passing an eval. Not when you’ve watched a video.

01 Beginner

Personal portfolio / landing page

The core loop: prompt → generate → preview → deploy, and reading diffs.

done → Live on a public URL (Vercel / Cloudflare Pages).

05 Intermediate

Full-stack CRUD app with auth + DB

Schema design, env vars & secrets, a real backend.

done → Deployed, you can sign in and your data survives a refresh.

07 Intermediate

Your first MCP server

How agents get tools — build one and test it with MCP Inspector.

done → Your editor’s agent can call your tool.

12 Advanced

Spec-driven multi-agent capstone

Decomposition, parallel agents against a spec, verify-then-ship.

done → A real multi-feature product, built to acceptance criteria and live.

All 12 projects →

The good stuff, curated

The docs, courses, and videos actually worth your time — pulled from what works, not what ranks. No affiliate padding.

Building Effective Agents Anthropic The reference mental model for agents vs workflows. Read this first.
Effective context engineering for AI agents Anthropic How to feed an agent the right context — the skill that transfers across every tool.
Claude Code best practices Anthropic Official playbook for the explore → plan → code → commit loop.
Codex CLI features OpenAI Approval/sandbox modes, AGENTS.md, and headless codex exec — the canonical source.
opencode docs SST The model-agnostic terminal agent. Good for learning that "the agent" and "the model" are separable.
Windsurf Cascade docs Windsurf Agentic edits with checkpoints — teach rollback discipline.

The full library →

Build log

Real builds, broken down — with the numbers. The receipts behind everything taught here.

Jul 1, 2026

I rebuilt this hub in an afternoon with an agent — here's the exact loop

The channel is back, as a full learning hub. The first build log is the site itself: how it got built, the loop that built it, and why 'shipped' is the only metric that counts here.

Read it →

Work with me

Want it shipped, not taught?

Some teams don’t want a course — they want a working AI agent in production by next quarter. That’s the day job. Fixed-scope audit, build sprint, or fractional retainer.

See how it works →

Free · the newsletter

Get the build logs + the Mirror System

One email a week: a real build broken down, plus a working reference agent you can clone. No fluff, unsubscribe anytime.