HomeEnterprise
Enterprise · deployment, identity & governance

Your data path. Your control.

Choose where inference runs — Medrecords is never required to be in your PHI data path. Three deployment models, one platform, identical outputs.

Deployment · AI Enablement your tenancy
Inference runs in your AWS / Azure accountin-account
Model keys — owned by youyour keys
PHI path — never transits Medrecordsisolated
Audit trail — HMAC-signed, append-onlysigned
SSO / SCIM — Okta, Entra ID, PingSAML 2.0
We ship the intelligence; you keep the data
Deployment models

Three ways to deploy. One platform.

Identical outputs and the same verification rigor — you choose where the records live and where the models run.

Not sure where to start?

Answer four quick questions — the plan that fits lights up below, with the trade-off spelled out.

1Who's deploying?
2What are you optimizing for?
3How strict is your PHI policy?
4Setup & procurement?

Answer above — the plan that fits lights up below, with the trade-off spelled out.

Plan 01 · if you're a firm or clinic

Self-Service

Pay per page in our HIPAA-eligible cloud — no subscription, no procurement.

PHI data path
✓ BAA SIGNEDAES-256 / TLS 1.2+ end to end
Model keys
✓ MANAGEDheld and rotated for you
Time to pilot
✓ START TODAY
Integration
✓ 3 PATHSupload portal · SFTP drop · API
Best for
Law firms, IME clinics, small TPAs
✓ PER-CASEvolume bundles available
Start with Self-Service
The differentiator
Plan 02 · if you're a carrier or large TPA

AI Enablement · BYO keys

Inference runs in your own AWS Bedrock / Azure OpenAI account.

PHI data path
✓ YOUR TENANCYrecords and model calls never transit Medrecords
Model keys
✓ YOUR KEYSyour BAA, your dashboards, your spend controls
Time to pilot
✓ 1 WEEKyour tenancy, wired in days
Integration
✓ VPC PEERINGplus every Self-Service path
Best for
Carriers & large TPAs with strict cloud-security policies
✓ $15K/YR + $5K SETUPcompute on your bill
Talk through AI Enablement
Plan 03 · if procurement says SaaS is out

On-Prem / Private Cloud

Full stack inside your infrastructure. Air-gap option.

PHI data path
✓ YOUR NETWORKnothing leaves it
Model keys
✓ LOCAL MODELSor models in your private cloud
Time to pilot
✓ ~4 WEEKSscoped with your infra team
Integration
✓ FULL APIclaims-system integration included
Best for
Regulated programs, self-insured funds
✓ $45K/YR + $5K SETUPuncapped, scoped on the call
Scope an on-prem deployment

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.

Identity & access

Everything procurement scans for.

Enterprise identity, provisioning, and access control — stated in the exact terms your security review checks against.

SAML 2.0 Single Sign-On — Okta, Entra ID, Ping
SCIM user provisioning & deprovisioning
Role-based access control — granular, per-matter permissions
Full user activity logs, exportable
Multi-factor authentication enforced
Configurable data retention & deletion policies
AI governance

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.

GOV-01

Scoped actions

The system operates from an explicit command whitelist — anything outside scope is structurally impossible, not just discouraged.

GOV-02

Approval-gated writes

Any action that leaves the platform — export, notification, external request — requires explicit human approval. Analysis is autonomous; action is governed.

GOV-03

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.

GOV-04

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.

Reliability & security

The spec sheet.

Built for portfolios, not sample files — and stated plainly, with nothing overstated.

Visit the Trust Center
SCALE10,000+ page portfolios — streaming ingestion, auto-resume on interruption, per-stage timeouts and retry queues. No case lost to a crash.
MODESFAST 2–5 min · STANDARD auto · THOROUGH 60–90 min / 2,000 pages
ENCRYPTIONAES-256 at rest · TLS 1.2+ in transit — every file, every environment.
COMPLIANCEHIPAA controls · SOC 2 audit-ready · BAA / DPA available on all cloud plans.
TRAININGYour records never train an AI model. Contractual — it's in the BAA — not a settings toggle.
Dedicated CSM

A named customer success manager, not a queue.

SLA-backed response

Response times in the contract, not the marketing.

Tailored onboarding

Adoption program mapped to your desks and lines.

Quarterly security review

A standing review with your security team.

Enterprise FAQ

What your security review will ask.

Can Medrecords run without our medical records leaving our cloud account?

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.

Do you sign a BAA?

Yes, on all cloud plans. On AI Enablement, your existing BAA with AWS or Microsoft governs inference; ours covers the platform.

Do you support on-premise deployment?

Yes — the full stack inside your infrastructure or private cloud, with an air-gap option for sovereignty requirements. Typical deployment runs about 4 weeks.

Is our data used to train your models?

Never. It's contractual — in the BAA — not a settings toggle.

How is this different from human-review medical record services?

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.

We're evaluating several review platforms — what's actually different here?

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.