Imbue is working on training its own large models optimized for reasoning, with over 100 billion parameters. The company's initial focus is on developing reasoning models for internal enterprise uses, specifically concentrating on agents that can code. Imbue's ultimate goal is to enable anyone to build their own AI agents. The company also emphasizes on making models explain their reasoning, which improves transparency and trust. The co-founders envision an ecosystem where different people can customize their own computer to what they want it to be.
Key takeaways:
- AI research lab Imbue, formerly known as Generally Intelligent, has achieved a $1 billion valuation and unicorn status after a $200 million funding round led by Astera Institute, NVIDIA, and other investors.
- Imbue focuses on building custom, reasoning AI agents and large language models optimized for reasoning abilities, which they believe is crucial for effective AI agents.
- The company is working on developing AI agents for everyday tasks like writing code or analyzing policy submissions, with a focus on making models explain their reasoning for improved transparency and trust.
- Imbue's business model is open, aiming to make its reasoning-focused models accessible to both businesses and individuals, and allowing others to build on top of their work.