Reliable Intelligence Everywhere

What: Composability

Composability is a term we use to refer to the seamless, end-to-end integration of discrete units of cognitive functionality. In a TAI world, over 99% of the consumers of AI will themselves be Code/Software. In addition to human language, AI models will natively understand Software types. This is in contrast to “Chatbots”, which are valuable for assistant/copilot applications, but insufficient to enable end-to-end automation. 

How: Doing the right task

Today’s pre-trained LLMs still contain vast latent intelligence waiting to be unlocked through better methods, data, and interfaces. In addition to Richard Sutton’s “Bitter Lesson” [2], we believe that scale on its own will be insufficient to unlock AI’s full capability for automation. Automation demands systems that can consistently choose what to do and when to do it. In this sense, the “right task” is not an isolated domain-specific objective, but the general decision-making process that sits behind every action an AI system takes.

1. Positive-Sum Games with Long Term People.  We believe in expanding total value rather than competing for fixed resources. We collaboratively discuss rather than argue. We do not form political factions, we do not engage in zero-sum thinking or prisoner's dilemma dynamics, and we do not selfishly optimize at the expense of the team. When the team wins, we all win. [3]
2. Thinking in Bets.  Good decision-making involves a lot of uncertainty. We like to clarify our individual beliefs (which will often differ) of the state of the world or predictions about the future, and choose paths that have the highest expected ROI for the mission. We prioritize long-term goals over short-term gains.
3. First Principles. We want our decisions to always stem from clear reasoning, grounded in our values and culture. We try to challenge assumptions, seek truth, and clearly communicate the “why” for any decisions. While we do care a lot about each other's emotions, we try not to let them cloud our decision-making processes.
4. Prioritize Learning.  We recognize that a key to success is learning as fast as possible about ourselves, our team, our processes, our product, the market, and everything else involved with accomplishing our mission. We recognize that there is more knowledge in the world that we can ever assimilate, and we are always hungry for more. We plan to update this document as we continue to learn.
5. Passion.  We love what we do. We love it so much that we never stop doing it. When we're out at lunch, or walking around the neighborhood, we're talking about our technology and what we're creating. We’re looking for people who share that passion and are similarly enthralled by the challenges we've taken on.
6. A working environment built for human beings.  Many companies will try to force their employees to leave their emotions at home. We believe that emotions are a core part of what makes us human, and what ultimately fuels our passion for the work we do. We are always trying to create an environment where we can make rational decisions as a group, without suppressing our feelings as individuals. We build machines; we don't try to be machines.