GraphiteGraphite by Augsidius

Augsidius API

The Augsidius API brings clinical intelligence into your product: adaptive care models, evidence-backed decision support, clinical documentation and patient-facing conversation — all grounded in the Augsidius Evidence Graph, a curated, continuously graded map of clinical knowledge (conditions, findings, investigations, drugs, procedures and therapies, and how they relate).

Every answer the API gives is traceable: clinical statements come with the concepts they are about, the evidence behind them and a certainty label. The API is designed to support clinicians and care teams, not to replace them.

What you can build

Use case Models Typical product
Adaptive care pathways aug-care-model, aug-care-plan Care-coordination tools, chronic-care programmes
Doctor decision support aug-cds, aug-evidence-qa, aug-differential, aug-evidence, aug-med-safety, aug-lab-interpret Clinician copilots, e-prescribing safety, lab portals
Clinical documentation aug-scribe, aug-extract, aug-coding, aug-summary, aug-deid Ambient scribes, EMR note structuring, RCM / coding
Patient-facing experiences aug-interview, aug-triage, aug-patient-explain Symptom checkers, pre-visit intake, WhatsApp follow-up
EMR integration aug-cds (CDS Hooks card format) In-workflow alerts inside your EMR

How it fits together

  1. Observations in. You send what you know about a patient — free text, a transcript, structured fields, or a mix. The API turns it into clinical observations: findings, history, measurements, demographics, each present or ruled out.
  2. Reasoning on the evidence graph. Observations are matched against the Augsidius Evidence Graph to work out what is likely, what is dangerous, and what is still missing.
  3. Typed results out. Answers with citations, or streams of typed JSON lines — cards, conditions, next best actions, coded entities — that your software can act on. See Response types.
  4. Feedback back. When a clinician accepts or overrides a suggestion, tell us. It improves the evidence for everyone.

Familiar interface

The API follows the widely used chat-completions format, so any OpenAI-compatible SDK works by changing the base URL and key. Augsidius-specific options travel in one extra augsidius object.

Python
from openai import OpenAI
client = OpenAI(base_url="https://api.augsidius.care/v1", api_key="aug_test_...")
r = client.chat.completions.create(
    model="aug-evidence-qa",
    messages=[{"role": "user", "content": "Can metformin be continued at eGFR 25?"}],
)
print(r.choices[0].message.content)

Availability

Status Meaning
Live Generally available, covered by your plan's SLA
Preview Available to all plans for evaluation; response shapes may still change with notice
Coming soon Documented so you can design your integration; calls return 501 not_available until launch

Four models are in Preview: aug-evidence-qa, aug-care-model, aug-cds and aug-extract. The others are Coming soon — see Models and the Changelog.

Augsidius is clinical decision support. Outputs must be reviewed by a qualified professional before they influence care. See Safety, privacy & intended use.

Next steps