Former OpenAI Researcher Jerry Tworek Launches Core Automation

Jerry Tworek unveiled a new AI lab called Core Automation, stating his aim to build the 'most automated AI lab,' one that automates even its own research.

The company explained that rather than simply chasing bigger models and more data, it will develop new learning algorithms that go beyond pre-training and reinforcement learning, along with architectures that scale better than transformers.

• The core goal is a structure where a small team and powerful AI agents perform work that entire organizations used to do.
• The team consists of experts in frontier models, optimization, and systems engineering.
• Tworek left OpenAI in January 2026, believing it would be difficult to continue this kind of fundamental research there.
• Core Automation joins the Neo Labs trend led by OpenAI alumni, alongside labs like Thinking Machines Lab and Safe Superintelligence.