For CROs
Feasibility you can put in the bid — and defend in the review.
Win more bids with cited, auditable feasibility in hours — white-labelled under your brand, run in your own cloud, and called straight from your bid pipeline. When a sponsor challenges a number, you point to its public source and an audit hash.
White-label, dedicated-cluster, volume API.
The problem
A lost bid shouldn't cost you six weeks and six figures.
Feasibility done the old way is fine when you win and a write-off when you don't. You need a verdict fast enough to build into the bid, and cited well enough to survive a sponsor's scrutiny.
What you get
Built for how a CRO actually wins and delivers work.
White-label verdicts
Every cited verdict, audit trail, and export carries your brand — not ours. Hand it to a sponsor as your own work product.
Dedicated-cluster mode
Run the full data path in your own AWS or Azure account. Your protocols, your tenancy, your keys — the public control plane is never in the auth path.
Volume API & MCP
Call feasibility straight from your bid pipeline. Submit a portfolio of protocols in one batch and pull cited verdicts back programmatically.
Per-program isolation
Each sponsor program is partitioned. Runs, citations, and audit trails stay separated, with a runtime guard that refuses cross-program reads.
Part 11 audit export
One click pulls a tamper-evident, hash-chained audit bundle per study — the evidence your QA team and the sponsor need to trust the number.
Hours, not weeks
A six-week, six-figure feasibility cycle becomes a same-day cited verdict — fast enough to build into the bid, not after you have lost it.
Bid-defense workflow
Protocol in, defensible verdict in the deck.
Drop in the sponsor protocol
Upload the protocol PDF — or POST it to the API. The engine extracts indication, phase, endpoints, and eligibility automatically.
Ground every signal in public data
PTRS, site density, budget envelope, and cohort reach are computed against ClinicalTrials.gov, AACT, PubMed, FDA, EMA, Open Targets, and ChEMBL — each figure cited to its source record.
Defend the number in the bid
Drop the cited verdict into your bid-defense deck under your brand. When the sponsor challenges a site-density or enrollment estimate, point to the citation and the audit hash.
For your QA & security team
The questions procurement asks first.
Where does our data live, and who can see it?
In dedicated-cluster mode the entire data path runs in your own cloud account. The public control plane is not in your auth path and cannot read tenant data — that boundary is enforced at the VPC and again at the data-tier client.
How are sponsor programs kept separate?
Each program is partitioned by tenant ID, enforced at the query layer and re-checked by a runtime isolation guard that raises on any cross-tenant access. Dedicated clusters add a full account-level boundary on top.
Do our protocols ever reach a third-party model?
No. Inference runs on a self-hosted Qwen fleet on our own (or your own) GPUs. Protocols are never sent to a third-party API, and we never train a model on your data.
What can we hand to a sponsor or auditor?
A per-study, hash-chained audit bundle that records every action, actor, timestamp, payload hash, and reason code — exportable and anchored for tamper-evidence. It is designed against 21 CFR Part 11 and EMA Annex 11 controls.
What is the compliance posture today?
The Part 11 audit chain, HIPAA and GDPR design posture, and per-tenant isolation are implemented. SOC 2 Type II is in progress — see the Trust page for the current, honestly-labelled status of each control.
Volume API & batch submission
Decision-support only — not a medical device
Put a cited verdict in your next bid.
Book a demo and we'll run a real protocol through the platform — white-labelled, cited, and on a Part 11 audit trail you can defend in front of a sponsor.