How it works
Protocol in. Cited verdict out. In hours.
No black-box scoring, no weeks waiting on a vendor. Hand the engine a protocol and get a verdict where every number links to its public record — ready to verify, defend, and act on today.
Three deterministic steps: parse → ground → cited verdict.
How it works
Protocol in. Cited verdict out. In hours.
Ingest
Upload your protocol PDF. A self-hosted LLM extracts the endpoints, eligibility criteria, and trial design — your document never leaves our infrastructure for a third-party model.
Pull public data + score
We query ClinicalTrials.gov, PubMed, FDA, EMA, Open Targets, AACT, and ChEMBL, then run deterministic scorers for PTRS, budget, site density, and cohort reachability.
Cited verdict
You get a verdict where every number carries its source plus a tamper-evident Part 11 hash. Export the whole thing for your board or the FDA.
The three steps, in detail
What happens between upload and verdict
Each step is built so a clinical lead and a data team can both trace the result end to end.
You hand us the protocol
Upload the protocol PDF or paste a synopsis. The engine extracts the indication, intervention, target, phase, endpoints, and eligibility criteria — the structured handles every downstream scorer needs. Language models are used here for document understanding only; they never invent the numbers.
We ground every signal in public data
Each parsed handle is resolved against the seven public sources — ClinicalTrials.gov and AACT for site density and enrollment, PubMed for indication baselines, FDA and EMA for regulatory precedent, Open Targets for PTRS, and ChEMBL for molecule risk. The engine pins the exact record and snapshot date it reads from.
Deterministic scorers produce a cited verdict
Feasibility, PTRS, budget envelope, site density, and cohort reachability are computed by deterministic scorers — the same inputs and the same source snapshot produce the same verdict, every time. The whole run is written to a tamper-evident, hash-chained audit trail aligned to 21 CFR Part 11.
Why cited matters
A verdict you can hand to a board, a CRO, or a regulator
The output is not an opinion — it is a defensible argument.
Every figure in the verdict — PTRS, site density, enrollment velocity, budget envelope, cohort reachability — carries a citation back to the exact public record it was derived from, along with the query and the snapshot date. Open the citation on any number and you see the source row, so your team can confirm the claim independently rather than trusting a model.
Because the scorers are deterministic and every run lands on a hash-chained audit trail, a verdict produced today can be reproduced and re-audited months from now. That is the difference between a number you believe and a number you can defend.
See a verdict you can actually check.
Send us a protocol. We'll return a fully cited feasibility verdict you can trace, line by line, back to public data — yours to white-label, defend in a bid, or hand to a regulator.