Three ways to deploy. One platform.
Identical outputs and the same verification rigor — you choose where the records live and where the models run.
Answer four quick questions — the plan that fits lights up below, with the trade-off spelled out.
Answer above — the plan that fits lights up below, with the trade-off spelled out.
Self-Service
Pay per page in our HIPAA-eligible cloud — no subscription, no procurement.
AI Enablement · BYO keys
Inference runs in your own AWS Bedrock / Azure OpenAI account.
On-Prem / Private Cloud
Full stack inside your infrastructure. Air-gap option.
AI Enablement is the plan nobody else offers. Most vendors require your medical records to transit their cloud. With AI Enablement, extraction and verification run against models in your own cloud account — your BAA with AWS or Microsoft, your encryption keys, your audit logs. We ship the intelligence; you keep the data.
Everything procurement scans for.
Enterprise identity, provisioning, and access control — stated in the exact terms your security review checks against.
Autonomy you can take to your risk committee.
The analysis is autonomous and machine-verified. The actions are governed. Your experts make the calls — always.
Scoped actions
The system operates from an explicit command whitelist — anything outside scope is structurally impossible, not just discouraged.
Approval-gated writes
Any action that leaves the platform — export, notification, external request — requires explicit human approval. Analysis is autonomous; action is governed.
Tamper-evident audit trail
Every processing step, access event, and agent action is logged in an HMAC-signed, append-only trail — auditable by your team or a third party.
Explainable outputs
Every conclusion traces to source pages and a visible reasoning chain. No black-box verdicts. Your data is never used for model training without written consent.
The spec sheet.
Built for portfolios, not sample files — and stated plainly, with nothing overstated.
Visit the Trust CenterA named customer success manager, not a queue.
Response times in the contract, not the marketing.
Adoption program mapped to your desks and lines.
A standing review with your security team.
What your security review will ask.
Yes. On the AI Enablement plan, extraction and verification run against models in your own AWS Bedrock or Azure OpenAI account — your tenancy, your keys, your audit logs. Medrecords never touches PHI.
Yes, on all cloud plans. On AI Enablement, your existing BAA with AWS or Microsoft governs inference; ours covers the platform.
Yes — the full stack inside your infrastructure or private cloud, with an air-gap option for sovereignty requirements. Typical deployment runs about 4 weeks.
Never. It's contractual — in the BAA — not a settings toggle.
Human-QA vendors sample. Medrecords machine-verifies every extraction on every page — deterministic citation verification plus an independent cross-model check — the same rigor at page 1 and page 10,000. Your experts still make the calls.
Two things most alternatives can't offer: verification that runs on every page rather than a sampled human QA pass, and deployment flexibility — including inference inside your own cloud tenancy or fully on-prem.
Deploy it your way. Verify everything.
Three deployment models, one platform, the same verification rigor on every page.