What we build, and why we build it this way
SetMeld was founded on an observation that anyone who has worked inside a large organization will recognize: the hard part of enterprise AI is the data. What the model needs lives in forty systems that do not agree, and unifying them has always required a small army.
What we believe
Integration should be infrastructure you own. The knowledge of how your systems fit together should be an asset your organization keeps after the contract ends. Every design SetMeld generates is an artifact you can read, review and keep.
An answer without provenance is a rumor. A number that cannot be traced back to the system that produced it cannot carry a decision that matters. Attribution is a property of the graph.
The model should come to the data. The organizations with the most valuable data are exactly the ones that cannot send it anywhere. Running entirely inside a customer’s network, with a reasoning model they choose and host, is the default.
AI should ask when it is unsure. An assistant that quietly invents a mapping is worse than no assistant. SetMeld Composer narrates what it is doing and stops to ask the person who actually knows.
Who we are
SetMeld is built by a team that has delivered data systems for the University of Oxford, MIT, Microsoft and the Scottish Government, including work alongside Sir Tim Berners-Lee, inventor of the Web.
We are a small company working with a small number of organizations whose data problems are unusually hard. That is deliberate, and it is why the first conversation is with someone who can answer technical questions directly.
Where the team delivered data systems before SetMeld
Incorporated
SetMeld, Inc., Delaware, United States
Working with
Research institutions, regulated enterprises and public bodies in the UK, EU and US
Get in touch
Tell us about your data problem
The difficult ones are the ones we want to hear about.