We have been quietly building a new type of general frontier model since mid-2024. Current generation LLMs are optimized for chatting with humans via algorithms such as instruction-tuning and RLHF. That algorithm was pioneered by our co-founder Diogo Almeida with InstructGPT as an early researcher at OpenAI, and was the algorithm used to train ChatGPT. It has produced models that make incredible chat bot assistants, yet are too fundamentally flawed to automate large amounts of real work.

We have created the first Large Programmable Model (LPM), a new type of general model that is directly optimized for interfacing with software, algorithms, data, and decision making. Its high reliability and programmability enables the foundation for creating AI systems and agents that will transform the economy.

Co-founders

Diogo Almeida, CEO

Diogo Almeida, CEO

Erik Gafni, CTO

Erik Gafni, CTO

Sasha Sheng, COO

Sasha Sheng, COO

Technical Staff

 Brian Basham

Brian Basham

   Elise Liu

Elise Liu

  Clayton Ward

Clayton Ward

   Jeff Lin

Jeff Lin

 Val Chocklett

Val Chocklett

 Noam Samuel

Noam Samuel

  ** Ted Kalaw**

** Ted Kalaw**

   Brian Smiley

Brian Smiley

You?

You?

Join Our Team

What Makes Us Exciting

Ideal Candidate

Screenshot 2025-02-17 at 2.38.47โ€ฏPM.png