Prompt, Delegate, Automate

What You'll Learn

PDA started as a framework for remote workers using ChatGPT and Zapier. That version still works. But Claude Code changes what's possible, and it changes the approach. You're not crafting chat prompts anymore — you're building a system that knows how you work. You're not just asking AI to help — you're delegating to agents that can read your files, run code, and make decisions. You're not setting up Zapier flows — you're writing skills that run from the terminal.

• The new prompting — Why "prompt engineering" missed the point. CLAUDE.md and SOUL.md as persistent context. How to stop re-explaining yourself every session.
• Delegation with teeth — Skills, subagents, and autonomous agents. The difference between asking AI to help and giving it the tools to act. What can be delegated and what shouldn't be.
• Automation that compounds — The morning-brief skill, the content-pipeline, mail-triage. How to turn a repeated workflow into a single command, and why the investment pays off after the first week.
• The PDA stack — How the three pillars work together. Where each of your current workflows sits on the manual-to-automated spectrum, and which ones to move next.
• The bigger picture — How PDA connects to your Obsidian vault, daily notes, and the AI co-operating system pattern. The compound effect of getting the right things off your plate.

This course is built from real workflows. The skills referenced — morning-brief, content-pipeline, mail-triage — exist and run in production. If something sounds abstract, it's because I haven't found the right concrete example yet, not because the concept is theoretical.

Time commitment: About 2.5 hours total. Read it straight through or use it as reference — each module stands on its own.