← back
Ralph Loops: Build Dumb AI Loops That Ship — Chris Parsons, Cherrypick
Takeaway
A dumb-but-persistent agent loop ('Ralph') beats complex orchestration for getting real work shipped, in code and beyond.
Summary
- Hands-on two-hour workshop teaching 'Ralph Loops' — deliberately dumb, continuously-running AI loops (Claude Code/Codex/Cursor) that grind on tasks 24/7 instead of fancy multi-agent orchestration.
- Parsons reports running Ralph Loops continuously for emails, calendar, content, client work, and code — claiming most of his daily output now flows through them.
- Workshop builds a working Pomodoro timer from scratch using a Ralph Loop so attendees leave with a runnable pattern transferable to real work.
- Frames the technique as the natural successor to agile coaching: simple, persistent feedback loops beat elaborate orchestration.
agentsclaude-codeworkflow
Original description
Dumb loops beat clever workflows. Most teams building with AI agents reach for multi-agent orchestration, planning graphs, and elaborate tool chains. Then they spend months debugging them. A single loop that processes one ticket at a time, evaluates its own output, and improves on the next run will outperform all of it. In this hands-on workshop you will build three things. First, a working Ralph Loop that processes real tickets end-to-end. Second, a synthetic feedback loop so you can test and iterate locally without waiting on production data. Third, a self-improving cycle where the loop's output quality gets better with every run without you touching the prompt. Speaker info: - https://x.com/chrismdp - https://www.linkedin.com/in/chrisparsons/ - https://github.com/chrismdp Timestamps: 0:00 Introduction to the workshop and Ralph Loops 6:12 How AI agents work on loops 10:06 Using loops for software development 10:43 Live coding: Building a Pomodoro timer 12:13 Explaining the ticket system 15:01 Implementing the first ticket (Status command) 17:37 The simplest Ralph Loop: A while loop 22:12 Next steps and taking the concept to the next level 28:19 Implementing further tickets using TDD principles 40:33 Advanced feature: The 'loop' command in Claude Code 49:39 Structuring and managing Ralph Loops 55:33 Using sub-agents for better validation 59:39 The 'startup' skill and ambitious automation 1:04:10 Real-world application: Worker and morning loops 1:09:19 Q&A: Determining when to stop open-ended tasks 1:13:31 VCP command and project context 1:15:52 Q&A: Handling context rot 1:17:01 Q&A: Reviewing sessions and verification 1:22:38 Q&A: Versioning prompts and skills 1:27:12 Q&A: Knowledge management and system organization 1:38:34 Closing, feedback, and final questions