Loop Engineering
Published by O'Reilly Media, Inc.
Building Autonomous AI Coding Agents That Ship While You Sleep
What you’ll learn and how you can apply it
- Master the loop engineering mental model (ReAct lineage, trigger types)
- Learn to identify processes worth automating using loop engineering
- Deploy production-ready loops, including doc sweeps, error-fixing, and test-coverage
- Design and implement custom automation loops with appropriate stop conditions and budgets
- Apply safety guardrails, deterministic checks, and human-in-the-loop checkpoints for enterprise-grade autonomous coding agents
Course description
Prompting is becoming the bottleneck. The next leap in developer productivity is designing systems that prompt, evaluate, and iterate on their own. This is where loop engineering comes in. Loop engineering is the practice of creating autonomous agent workflows that operate in a continuous cycle of reasoning, acting, observing results, and repeating until a defined goal is achieved. Join Harshit Tyagi to learn how to move beyond one-off prompts and adopt the ReAct (Reason, Act, Observe, Repeat) mindset, design effective triggers and measurable goals, and build loops that know when to stop. You’ll explore what makes a good loop, how to distinguish verifiable outcomes from subjective evaluation, and how to identify the kinds of repetitive development and operational tasks that are ideal candidates for automation.
Through hands-on examples, you’ll dissect seven production-ready loops for tasks such as documentation maintenance, log-driven bug fixing, SEO and GEO visibility improvements, and automated test coverage. You’ll learn how to implement and manage loops using tools such as Claude Code and Codex, isolate autonomous work with Git worktrees, and integrate agent outputs into existing workflows through pull requests and team notifications. The course also covers the practical realities of running agents at scale, including token costs, budgeting, safety guardrails, human oversight, and enterprise governance patterns. By the end, you’ll design and launch your own custom loop, complete with triggers, goals, stop conditions, and budget controls, giving you a repeatable framework for safely deploying autonomous agents in real-world development environments.
This live event is for you because...
- You’re a software engineer, AI/platform engineer, technical PM, engineering manager, or indie builder who’s already using an AI coding assistant.
Prerequisites
- Cursor, Codex, or Claude Code installed on your machine
- Comfort with an AI coding agent (Claude Code, Codex, or similar) and a Git workflow
Schedule
The time frames are only estimates and may vary according to how the class is progressing.
The loop mindset (20 minutes)
- Presentation: Understanding the shift from manual prompt engineering to autonomous loop engineering using the ReAct (Reason, Act, Observe, Repeat) framework; the two-ingredient model (trigger and goal) and differentiating between manual, scheduled, and goal-based triggers; criteria for verifiable goals versus LLM-as-judge evaluation to ensure loops have clear stop conditions
Good use cases and best practices for setting up loops (45 minutes)
- Presentation: Seven production-ready loops, including documentation sweeps, log error-fixing, SEO/GEO visibility, and test-coverage automation; deconstructing each loop with specific copy-paste prompts and expected iteration counts for immediate implementation; analyzing “good goal versus bad goal” design
- Break
Tooling and setup (25 minutes)
- Presentation: Implementing operational loops using tools like Claude Code and Codex; automation, scheduling, and goal-setting commands; using git worktrees to isolate parallel agents to ensure a safe, non-destructive development environment; wiring agent outputs back into human workflows (Slack pings, PRs) to maintain visibility and control over autonomous actions
Cost, safety, and governance (20 minutes)
- Presentation: The economic reality of token usage and budgets to prevent runaway costs in automated loops; implementing safety guardrails, deterministic checks, and human-in-the-loop checkpoints to mitigate risks; PR-gated merges and permission layers for creating an enterprise-grade framework for responsible autonomous agent adoption
Design your own loop and capstone (10 minutes)
- Presentation: Creating a custom loop from scratch using the provided worksheet, covering triggers, goals, stop conditions, and budgets
- Hands-on exercise: Launch one custom loop for your own code base and document the iteration results
Your Instructor
Harshit Tyagi
Harshit Tyagi is an AI consultant, founder, and educator focused on building AI systems that create real business value. As the founder of Agentwise, he helps teams move beyond prototypes by designing agent systems, automation platforms, and production-grade AI applications that actually ship and scale. He is also the founder of glued.me, a marketing SaaS platform built to automate the full performance marketing stack—from campaigns and creative to bidding and reporting through agent-driven workflows.
Beyond building products, Harshit teaches AI to a global audience through platforms such as LinkedIn Learning, O’Reilly, and YouTube, where he is known for turning complex ideas into clear, practical, example-driven explanations. His work sits at the intersection of AI strategy, product execution, and education, with a strong belief that the best AI is not the kind that looks impressive in a demo, but the kind that improves day-to-day operations and moves real metrics.