How to Ship Your First Pull Request as a PM

Key Takeaways

• Garry Tan ships 100 PRs/week; repo gained 65K GitHub stars.
• Anthropic PMs generate 20‑30 AI‑written PRs daily.
• OpenAI Codex product built with 1,500 merged PRs, no hand‑coded lines.
• PM‑focused planning moves specs into Git markdown, cutting doc overhead.
• Shipping code lets PMs test ideas in hours, not quarters.

Pulse Analysis

The rise of AI‑assisted development tools is blurring the traditional lines between product management and engineering. High‑profile figures such as Garry Tan and Boris Cherny demonstrate that product leaders can now generate and merge pull requests at a scale once reserved for seasoned engineers. By leveraging large‑language models like Claude and Codex, PMs can produce functional code snippets, run automated tests, and submit PRs that pass initial reviews, dramatically reducing the time from concept to production.

This shift is not merely a novelty; it addresses a chronic bottleneck in software delivery. Engineering teams often spend a disproportionate amount of time writing boilerplate or low‑impact features, while strategic work stalls awaiting specifications. Moving planning artifacts into Git—using markdown files like PLANNING.md and CLAUDE.md—places the product brief directly alongside the codebase, ensuring version‑controlled, searchable, and auditable specifications. The result is a tighter feedback loop: PMs can A/B test copy changes in a single afternoon, iterate on UI tweaks without waiting for a full sprint, and validate hypotheses before committing significant engineering resources.

For organizations, the business impact is measurable. Companies adopting this model report up to a 200% increase in code output per engineer, as routine changes are off‑loaded to PMs or AI agents. The reduced cycle time translates into faster time‑to‑market, higher customer satisfaction, and lower opportunity cost for high‑risk projects. As the practice matures, we can expect broader adoption of open‑source planning systems, standardized PR skill files, and cross‑functional playbooks that empower product teams to own end‑to‑end delivery while preserving engineering focus on complex, high‑value problems.