Product
One question. Your answer in seconds.
Thousands of consultation notes, read and matched to your protocol while you're still designing it.
How it works in three steps
Pick your protocol
And its eligibility criteria: species, breed, age, diagnosis, bloods, treatment history.
Search the Ulysse database
Every record already standardised, de-identified and ready to query.
Get the answer
Against real clinical records: how many animals qualify, which practices and cities, which individual patients.
Live feasibility
Switch protocol and watch the answer change. Every number comes from a consultation note that arrived as an unstructured document.
How many
Eligible patients in the network
Where
Practices, cities and groups holding them
Who
De-identified patients, with the diagnosis as the vet wrote it
The demo runs on 320 synthetic consultation notes from four practice systems, in English and French. Open the full console
How Ulysse works: from PDFs to cohort
One engine reads, structures and matches. Nothing is copied out to a third party.
Stage 1
Records in
Notes from any practice system, any layout, any language. PDFs, scans and exports — no template needed per system.
Stage 2
Read and structured
Each document is read automatically. Breed, age, weight, diagnosis, medication and bloods are pulled out as fields.
Stage 3
Standardised and de-identified
“Renal ca” and “carcinome du rein” become one condition. Owner identity never leaves the practice.
Stage 4
Matched to your protocol
Every patient is scored against your criteria: eligible, near miss, or out — with the failing criterion named.
Measured, not claimed
Every extracted field is graded against ground truth.
Separation built in
Sponsors and practices see different views of the same records, enforced in the data itself.
Synthetic demo
No real animal, owner, practice or sponsor appears anywhere on this site.