Week 1 of 8 · Use AI

The AI-powered development workflow

Learn by building. Engineers build real features from Day 1 using AI-powered tools and workflows, and end the week faster developers using agentic programming.

What the week builds toward

WorkBot and its public site scaffolded, 4 or more features shipped in parallel across them on git worktrees, a full SDLC loop from Linear ticket to deploy at 10 PRs an hour, and 2 custom MCP servers connected to Claude Code.

The 5 days

  1. Day 1

    Dev environment and Claude Code fundamentals

    Set up the AI development environment and run the full code-to-deploy loop until you can sustain 10 pull requests an hour.

    You ship: A Next.js project scaffolded with Claude Code, and the PR loop run to 10 an hour.

  2. Day 2

    Claude Code: skills and sub-agents

    Extend Claude Code with custom skills and parallel sub-agents, and build a 3-agent pull request review.

    You ship: 3 custom skills and a 3-agent PR review.

  3. Day 3

    The AI SDLC: feature development workflow

    Run the full AI SDLC from Linear ticket to deployment through MCP integrations, and learn to write specs that AI can implement from.

    You ship: A Linear backlog shipped through the full SDLC at 10 PRs an hour.

  4. Day 4

    Multi-project and multi-feature workflow

    Run several Claude Code instances in parallel on git worktrees and ship 4 or more features in a single session.

    You ship: 4 or more features shipped in parallel, with throughput measured.

  5. Day 5

    MCP servers and complete feature development

    Build custom MCP servers in TypeScript, and finish a feature with tests, analytics and monitoring.

    You ship: A database MCP server and a PostHog MCP server, both used from Claude Code.