The intelligence layer for businesses that run on unstructured records.
Troml sits underneath the systems you already have. It takes what your organisation generates — correspondence, documents, minutes, scanned archives, regulation — and makes it answerable rather than searchable.
Ask what is true right now and the answer is assembled across every source that bears on it, with a citation for each claim. Not a ranked list of files that mention the words.
A document retrieval system answers “where is the file that mentions X?” Troml answers “what is true about X right now, and how do you know?”
Measured, dated and graded.
Every figure below is measured in a live production system or on a public benchmark using its own unmodified scoring harness. Nothing here is a projection.
On the benchmark. That run was self-administered using the benchmark’s own unmodified scoring harness. An independent re-grade by the organiser is outstanding, so we quote the score and not a ranking.
What we will not claim.
These are our own findings, from our own audits. You would find them in due diligence. We would rather you found them here.
Self-administered on the official harness. The independent re-grade has not landed.
Our original token-reduction study was invalidated by our own audit — it measured about 5% of true cost. The re-measurement is unpublished, so the claim is absent rather than softened.
It is on file. It is also why the numbers that survived later measurement are worth something.
Substrate primitives ran unchanged in four of five cases in a second vertical — but that vertical is a pilot in the same jurisdiction.
We upgraded a core component to a materially better one and the whole system got worse — 74.4 against 76.9. It fixed 16 questions and broke 31. We publish the losses because they are how we know the wins are real.
Bring a corpus, not a brief.
Point us at a slice of your real data — one business unit, one document class, one year. Real records with real ambiguity in them; curated sets make every system look good and tell you nothing.
We run the domain probes and show you, against your own records, what can already be answered and where your corpus has gaps. You get the gap list whether or not you proceed.
The full technical brief — architecture, ingestion discipline, governance, deployment and a limitations section — is available on request.