University Accreditation Lookup
It's not a lookup problem. It's a naming problem.
Verifying a university's accreditation sounds like a lookup. In practice, it starts with a naming problem. A school name arrives from an applicant or a third party in whatever form they happened to type it, and the real institution of record is often a step removed from what was submitted. "Fuqua School of Business" has to be recognized as Duke University before any accreditation check can even begin. Multiply that across thousands of name variants, abbreviations, and sub-schools, and the lookup was the easy part. Getting the name right upon intake was the challenge.
A human reviewer handles this by knowing, or looking up, which approved source to trust and how to read it. That works, and for a clean list of well-known schools it can work quickly. It gets slower and less consistent as the names get messier, the shift gets longer, and the reviewer has done the same search a thousand times that week.
What we built
TNDM's accreditation agent normalizes the submitted name, resolves it to the actual institution, and checks it against the same approved sources a human reviewer would use: the federal DAPIP registry first, then CHEA and the relevant regional accreditor when needed. Every result comes back with a confidence score, a plain-language explanation of how that score was reached, and the specific source the answer relied on.
A lookup for "Fuqua School of Business" in North Carolina comes back resolved to Duke University, matched against DAPIP by exact name and state, with a confidence score of 95 and an explanation naming the federal registry as the source and noting why that source carries high authority. A lookup for a school that doesn't exist comes back with a confidence score of zero and an explanation that no valid accreditation record could be found — it just tells you it couldn't find one.
That second case is the one I actually care more about. If a tool can't admit when it doesn't know something, you can't really trust it with a compliance decision, and we spent as much effort teaching it to say "I don't know" as we did teaching it to find the right match.
Where it stands
This one is live in production today, not a prototype. It's part of what TNDM offers as a university accreditation agent, and the cost difference is real too — it's priced per execution, it runs well under what a human verification costs, and that gap widens as volume grows. But the number I'd actually stand behind isn't the savings percentage. It's that the agent runs the same approved sources, the same steps, every time, at 2:00 in the morning or 9:00 at night, and shows its work every single time.
© 2026 Wade Hudson. All rights reserved. Share the link or quote a short excerpt with credit, but please don’t republish this write-up without permission.