vs. raw LLM output
Deterministic, not probabilistic
An agent that guesses a DRG is a liability. Give it a tool that computes the answer the same way every time — and cite the rule version that produced it.
Quantara Health
Agents are starting to work claims, code charts, and appeal denials. Quantara is the ground truth they call: deterministic APIs for DRG grouping, claim validation, medical necessity, and Medicare pricing — versioned, auditable, built to be a tool.
REST + MCP · Deterministic · Versioned against every CMS release.
Agents will do the work of the revenue cycle. The rules they follow can't be vibes.
Language models are probabilistic; payment policy is law. When an agent codes a chart, scrubs a claim, or drafts an appeal, it needs a deterministic answer to "what does the rule say?" — not its best recollection of a PDF from training data.
Quantara ships payment policy as versioned code: every grouper release, code edit, and coverage determination as an endpoint an agent can call, a version you can pin, and an audit trail you can show. Agents do the work. Quantara keeps them honest.
vs. raw LLM output
An agent that guesses a DRG is a liability. Give it a tool that computes the answer the same way every time — and cite the rule version that produced it.
vs. legacy installs
The incumbents ship DLLs, desktop encoders, and quarterly update disks. Quantara is a POST request or an MCP tool — always current, nothing to deploy.
vs. PDFs and manuals
Pin a rule version like a dependency. Diff one grouper release against the next. Reprice last year's claims under this year's rules.
Small, composable, and callable by humans or agents — every endpoint as REST and as an MCP tool. Use one, or chain them into a pipeline.
POST /v1/group
MS-DRG assignment from ICD-10 diagnoses, procedures, and demographics. CMS version resolved from the discharge date, MCE edits included.
View docs →
POST /v1/validate
Code validity, MCE-style edits, POA and HAC checks — catch the denial before the claim leaves your system.
Spec coming soon
POST /v1/necessity
Check a service against NCD and LCD coverage policy for a given diagnosis, place of service, and date.
Spec coming soon
POST /v1/price
What Medicare would pay: IPPS and OPPS pricing with wage index, add-ons, and outlier logic itemized.
Spec coming soon
GET /v1/providers
NPI lookup and enrichment: taxonomy, enrollment, and practice data — clean, current, and joined for you.
Spec coming soon
POST /v1/rules
The small stuff that breaks claims: timely filing windows, modifier logic, frequency limits — as callable rules.
Spec coming soon
One request in, one defensible answer out. The CMS version is resolved from the discharge date, returned in the response, and reproducible forever. Try it live in the docs →
Request
curl https://api.quantarahealth.com/v1/group \
-H "Authorization: Bearer $QH_TOKEN" \
-d '{
"discharge_date": "2026-07-01",
"principal_diagnosis": "I21.4",
"secondary_diagnoses": ["E11.9"],
"secondary_diagnosis_poa": { "E11.9": "Y" },
"procedures": ["027034Z"],
"age": 67,
"sex": "F",
"discharge_status": "01"
}'
Response — 200
{
"request_id": "req_8f3a…",
"ms_drg": "322",
"title": "Percutaneous Cardiovascular
Procedures with Intraluminal
Device without MCC",
"mdc": "05",
"cms_version": "43.1",
"effective_date_range": {
"start": "2026-04-01",
"end": "2026-09-30"
},
"mce_edits": [],
"warnings": ["PYTHON_CMS_ALGORITHM_PARTIAL: …"],
"billable_units": 1
}
Quantara runs inside other products — and inside other agents. If you're building healthcare software, this is your payment-policy layer.
Coding copilots, claim-scrubbing agents, appeal writers: hand them Quantara as a tool and every action they take cites a rule version instead of a hallucination.
Show the DRG, the edits, and the coverage answer while the chart is still open — not three weeks later on a remit.
Validate codes, check necessity, and price the claim in one pipeline pass — without licensing a desktop encoder per seat.
New payers, TPAs, and risk startups get decades of payment policy as endpoints instead of headcount.
Run last year's claims through this year's grouper — or next year's proposed rule — and quantify the impact before it lands.
Stateless by design.
Requests are processed and answered — grouping and validation don't require us to keep your claims. Every response carries its rule version and an audit log, so when an agent acts on it, you can prove exactly what it knew and when. Your compliance team meets no surprises.
The basics, handled
A real API key, in thirty seconds.
Ready for production volumes, a BAA, or pricing? Book a call — trial keys upgrade in place.
Ready for production?
Already tried your free key? Book a call to talk production volumes, usage-based pricing, BAAs, and SLAs — or tell us what you're building and we'll map the integration with you.