In the mid-1990s, major internet companies raced to build the “total” web portal. Yahoo, Excite, Lycos, and AOL spent billions acquiring and assembling services to become the single destination for everything users needed online. They competed in what the media dubbed the “portal wars,” convinced the future belonged to whoever could consolidate the most comprehensive walled garden.

In 1998, Google made the opposite bet. While portals tried to be everything to everyone, Google built a data-driven recommendation layer: an algorithm that analyzed the distributed web and sent each query to its best destination. Their strategy won.

Today, intelligence is fragmenting. There is no single monolithic oracle everyone uses for everything, and there will not be: the diversification of models and agents is outpaced only by the diversification of tasks and data they are exposed to. As a result, the interaction space between models, agents, and their context has become as complex and opaque as the models themselves—rebuilt anew every three months, constantly evolving, nearly impossible to interpret manually.

Not Diamond is the data-driven recommendation layer for intelligence. Our research helps developers understand and improve agent behavior at scale, replacing scattershot manual processes with 100x improvements across quality, cost, and engineering efficiency. We believe infrastructure to enable a distributed future for AI will not only drive forward performance and reliability, but also make the field less monopolistic, more energy efficient, and more interpretable. We aim to achieve nothing less.

We’re a small team working to alter the course of the future. If you’d like to join us, we’d like to hear from you.

INVESTORS

Jeff Dean Google
Julien Chaumond Hugging Face
Ion Stoica Databricks, Anyscale
Akshay Kothari Notion
Arash Ferdowsi Dropbox
Guillermo Rauch Vercel
Olivier Pomel DataDog
Zack Kass Open AI
Lukas Biewald Weights & Biases
Jeff Weiner LinkedIn
Scott Belsky Adobe
Tom Preston-Werner Github
Dan Roth Oracle
Dwarak Rajagopal Snowflake
Amir Haghighat Baseten
Eoghan McCabe Intercom
Matias Woloski Auth0
John Kim PayPal
Paul Forster Indeed
Nadim Hossain Databricks
Amanpreet Singh Contextual
Chet Kapoor Datastax
Carl Rivera Shopify
Kiran Prasad LinkedIn
Neil Sequiera Defy
Dean Mai Myriad Venture Partners
Preetha Parthasarathy SAP
Caitlin Dullanty IBM
Adam Gartenberg 640 Oxford
Steven Woods Inovia Capital

notes on culture

We are a small, dedicated team that spikes on elite technical caliber, exceptional emotional intelligence, and long-term, impact-driven bets.

1 It’s easy to confuse our worth as people with the value of our work. We think it’s critical to separate the two—the value of our work is not correlated to our worth as people. Not only do we think this is true, but we also believe it frees us up to speak more honestly to where we are succeeding or failing as individuals or a team. Once the value of our work is no longer a vehicle for shoring up insecurities or ambitions in our egos, we can be more clear-sighted about what is actually the case. We value low egos, exceptional emotional intelligence, and deep mission alignment, and we all work to contribute to a psychologically safe, blame-free environment.

2 There are many reasons why people want to work on startups: financial success, autonomy, learning. But the truth is that the vast majority of startups fail. They are incredibly stressful, demoralizing, and statistically a very poor strategy for making money. For this reason, we believe it’s important to work on problems that are so important to us that we would feel proud to have worked on them even if we did fail. We’re working on Not Diamond because we believe that infrastructure for ensembling and optimizing diverse machine intelligences has the potential to produce enormous public good and create a safer future for humanity. That is a future worth failing for.

3 Because startups are so incredibly difficult, we believe it’s essential to constantly develop ourselves to be world-class. We don’t mean this rhetorically: we literally strive to be amongst the best in the world at the problems we’re solving. Whatever the nature of our work, we seek to do it exceptionally. Excellence extends not only to our technical execution, but also to our emotional intelligence, collaboration, and self-care. Importantly, we believe our greatest impact comes not from hours worked at the margin, but from exceptional decisions and exceptional relationships. Excellence is built over a lifetime of sustainable progress—we try to make some every day.

4 We want to work with people we feel comfortable building long-term relationships with, even as the focus of those relationships constantly evolve. Life is too short not to work with people you love. When we make a commitment to work together, we predicate that commitment on a long-term, high-autonomy bet in our collaborators. To hold ourselves accountable to this, we offer an investment in any future startup that a team member might start in the future. Making this offer forces us to ask whether we believe in somebody over the long term, and if we truly believe they have the skill, ambition, and creativity to wrest value from nothing.

5 We believe diversity is critical to the quality of our work. At the same time, a lack of diversity is equally important. Part of what makes the best teams effective is an uncommon density of people who think and act very similarly. We don’t want diversity in our commitment to psychological safety, impact, or excellence. Importantly, defining where we don’t want diversity allows us to then liberally maximize diversity outside of those bounds. In doing so, we remember that the areas where we don’t want diversity—our values—are not static. Rather than drifting into cultural practices unconsciously, we evolve our culture intentionally and collaboratively.

100x your AI dev cycles

Let the machine build the machine