Research acceleration: The view inside OpenAI
OpenAI researchers are increasingly relying on autonomous coding agents, signaling a massive shift in how software and models get built.
Simon Willison ยท Sep 7, 2026
Even the teams creating cutting-edge AI are changing how they write software. Internal updates from OpenAI reveal that their own researchers now lean heavily on automated coding helpers to speed up research, write scripts, and handle complex technical workflows.
As technologist Simon Willison noted, this reflects how "agentic engineering really took off" inside leading labs. When the engineers building the models delegate their daily coding tasks to AI agents, it confirms that the way software gets made has fundamentally evolved.
For anyone learning to build apps with modern AI tools, this is great validation. Success no longer depends on memorizing obscure syntax. Mastering the ability to guide agents, structure projects clearly, and test outputs puts you right in line with how top industry engineers work today.
Originally reported by Simon Willison
Read the originalMore news
Using Blender with coding agents on macOS
AI coding agents can now pilot 3D software like Blender directly, letting you generate custom 3D scenes and animations using simple prompts.
Source: Simon Willison
GPT 6 Astra now available on Vercel AI Gateway
Vercel has added OpenAI's GPT 6 Astra to its AI Gateway, giving builders access to a model designed specifically for long-running coding agents.
Source: Vercel / v0 News
Gemini 3.8 Flash now available on AI Gateway
Google's Gemini 3.8 Flash is now live on Vercel AI Gateway, offering fast, budget-friendly reasoning for tools like Cursor and Claude Code.
Source: Vercel / v0 News
Weekly vibe-coding tips โ coming soon
We're getting this ready. Leave your email and we'll let you know the moment it launches. No fluff, unsubscribe whenever.