Clinical researchers
Build cohorts and test hypotheses without the data-engineering lift.
Multi-vendor clinical data
Clinical data lives in silos. Tessera unifies it as one corpus, purpose-built for AI.
Bedside monitors stream high-frequency signal that is typically kept only weeks before it clears. The EHR holds what was done about it, on a different clock. CalmWave captured both and fused them into a single corpus, enriched with 40,000+ academic papers. Tessera opens it to the world, answerable in plain language.
Opening in small cohorts, in signup order. Secure your spot.
1.2M
de-identified patients
10.9B+
clinical and technical data points
40K+
linked academic papers
100
patient minimum per answer
Not synthetic. Not simulated. High-frequency signal recorded during real clinical care, then de-identified, structured, and fused: vital signs, physiological and technical device alarms, medications, billing codes, and more.
The silo problem
In pieces. And not for long.
The signal streams from the monitors, each vendor's system holding its own slice in its own format. The record of care, medications administered, diagnoses, interventions, lives in the EHR on its own clock. And the monitors keep the signal only a few weeks, long enough to snapshot into the chart, then it clears.
CalmWave captured all of it, then structured, normalized, and time-aligned the pieces into one corpus an LLM can actually reason over.
What it is
A queryable corpus of clinical signal
A structured corpus of high-frequency clinical signal (vitals, physiological and technical device alarms), normalized and time-aligned across streams, then fused with medications, interventions, diagnosis codes, and over 40,000 academic papers on the diagnoses in the data.
Ask a clinical question in plain language. Get an aggregate answer back in seconds, rather than a data-request ticket that takes a quarter.
Reachable over MCP, so an agent can query the corpus directly and get the same aggregate answer.
> Unit-stays with sustained heart rate over 140 during vasoactive infusions
returns a cohort, with the signal behind it, as an aggregate.
returns a cohort, with the signal behind it, as an aggregate.
> Medication administrations within the hour before a sustained SpO2 drop, grouped by drug class
returns an aggregate view across the corpus.
returns an aggregate view across the corpus.
> Technical alarms indicating sensor or hardware failure on bedside monitors, grouped by device type
returns an aggregate view of failure patterns across a multi-vendor fleet.
returns an aggregate view of failure patterns across a multi-vendor fleet.
See it run
The answer comes back in seconds.
Two live queries against the corpus, start to finish.
Who it is for
Built for the people who ask hard questions of data
Pharma, life sciences, and CROs
Real-world clinical signal at population scale.
Health-system analytics teams
One queryable source, no query team to stand up.
Algorithm developers
Structured signal to build and validate against.
Hardware and device engineers
Sensor and alarm behavior at fleet scale, across vendors, straight from the signal.
Clinical engineering leaders
Fleet-wide visibility into technical alarms and device failures, by unit and device type.
Trust
Built to be trusted
De-identified, aggregate-first
Every result aggregates a minimum of 100 patients. No record-level export.
Explainable and auditable
You can see how an answer was derived. No black box.
Privacy-preserving by design
Queries that would return fewer than 100 patients are rejected automatically, before any data comes back.
Early access
Hold your place.
Tessera opens in small cohorts of research and analytics teams. The first cohort is forming now, and access is granted in signup order.
No spam. One email when your cohort opens.
