OpenAI Team Shipped a Full Product Using Zero Human-Written Code

OpenAI Developers

Feb 11, 2026 · Updated Apr 25, 2026

OpenAI's engineering team built a real internal product entirely with Codex agents — 1,500 pull requests, roughly a million lines of code, zero manual coding. Engineering shifted from writing code to designing agent environments and feedback loops.

OpenAI published a case study on shipping a full internal product — used by hundreds of users — without a single manually-written line of code. Over five months, a 3-person team prompted Codex, OpenAI's coding agent, to open and merge 1,500 pull requests covering application logic, tests, CI, and observability. Codex was wired to Chrome DevTools Protocol for autonomous UI validation and agent-to-agent code review. A short AGENTS.md served as a table of contents pointing to a structured docs/ directory, keeping agent context manageable.

This shows agentic coding can sustain a full product lifecycle, not just individual tasks. Throughput reached 3.5 PRs per engineer per day and grew with headcount. Engineering roles shifted from writing code to specifying intent and designing environments where agents work reliably.

Read the full engineering post for patterns on context engineering, structured repo knowledge, and agent-to-agent review. Try Codex through the OpenAI platform.

View the full update on openai.com

OpenAI Developers

@OpenAIDevsFeb 11

📣 Shipping software with Codex without touching code. Here’s how a small team steering Codex opened and merged 1,500 pull requests to deliver a product used by hundreds of internal users with zero manual coding. https://t.co/2GaeX7We2n

203

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Keep reading

OpenAI Says Codex Can Already Work for Days, Easier Controls Coming

OpenAI says its Codex agentic coding platform can already work for days on hard tasks with small configuration tweaks. The team plans to release changes that make these long-running workflows easier for everyone, signaling a near-term move toward multi-day autonomous engineering sessions.

Vercel Shares Engineering Framework for Shipping Agent-Generated Code Safely

Guillermo RauchMar 31

Vercel Shares Engineering Framework for Shipping Agent-Generated Code Safely

Vercel released its internal guidance for agenting responsibly after shifting to a workflow where AI agents perform the majority of their coding. The framework moves beyond traditional CI testing to include executable guardrails and autonomous deployment rollbacks that contain the risk of AI-generated errors.

OpenAI Open Sources Auto-review to Automate Safety Checks for Codex Agents

Maja TrebaczMay 4

OpenAI Open Sources Auto-review to Automate Safety Checks for Codex Agents

OpenAI released the research and code for Auto-review, a secondary agent that handles permission requests for Codex without requiring human intervention. This architecture allows autonomous coding agents to perform sensitive tasks like network calls while maintaining safety oversight through a separate reasoning model.