The raw signal, as JSON

Authentication

Every request carries your key as a bearer token. Keys are stored hashed on our side, which means we cannot show you a key again after it is issued: keep it somewhere safe, and if you lose it we revoke it and issue another.

curl https://biodatica.com/api/v1/companies?area=oncology \
  -H "Authorization: Bearer YOUR_KEY"

Keys belong to an account, and they carry that account’s plan. A key on an account without API access is rejected rather than silently downgraded. A key can be revoked from your account at any time.

Rate limiting is 120 requests per minute, per key, and it is counted per key rather than per IP so that one customer’s backfill cannot throttle another customer sharing an egress address. Over the limit you get a 429 with a retry-after header. If your integration needs a higher ceiling, that is a conversation and not a support ticket.

Endpoints

Base URL https://biodatica.com/api/v1. Everything is JSON, and all timestamps are epoch milliseconds in UTC, matching the graph’s own storage.

{
  "data": {
    "company": { … },          // the shape below
    "events": [ … ],
    "grants": [ … ], "trials": [ … ],
    "patents": [ … ], "filings": [ … ],
    "link_evidence": [ … ]     // method + score behind every edge
  },
  "notice": "…"
}

What comes back

Every response carries the same envelope. data is the payload, page appears on the list endpoints, and notice is the standing disclosure that rides on the JSON as well as the pages. Errors replace all of it with a single error object.

{
  "data": [ … ],
  "page": { "limit": 50, "offset": 0, "count": 50, "has_more": true },
  "notice": "Informational only. Not investment advice. …"
}

// and on an error, at any status:
{ "error": { "code": "rate_limited", "message": "…" } }

Inside data, field names match the graph’s own schema, so what you read here is what the product reads internally. A company:

{
  "id": "…",
  "name": "…",
  "slug": "…",
  "aka": "…",                  // JSON array of aliases, or null
  "city": "…", "state": "…", "country": "US",
  "website": "…",              // or null
  "therapeutic_area": "oncology",
  "modality": "…",             // free text, or null
  "stage": "Phase I",
  "momentum": 84,              // 0–100
  "momentum_delta": 6,         // change since the last recompute
  "runway_months": 7,          // null = unknown. NEVER 0 for unknown.
  "last_raise_cents": 2600000000,
  "last_raise_at": 1750000000000,
  "ai_brief": "…",             // generated commentary, or null
  "ai_brief_at": 1750000000000,
  "confidence": 0.91,          // crosswalk confidence
  "curated": 1,                // 1 = a human confirmed the edges
  "updated_at": 1750000000000
}

An event:

{
  "id": "…",
  "company_id": "…",
  "type": "trial_registered",
  "weight": 22,
  "headline": "Phase I trial registered",
  "occurred_at": 1750000000000, // the date the FILER reported
  "source": "trials",
  "source_url": "https://clinicaltrials.gov/study/NCT…",
  "payload": "…"                // JSON detail
}

Two field notes worth reading twice. runway_months is null when it is unknown and 0 only when the model says the last disclosed raise is spent. Do not coalesce null to zero in your pipeline; they mean opposite things. And occurred_at is the filer’s date, not our ingest date, so a backfilled event can land in your feed with a timestamp in the past.

Enumerations

These are closed sets. We may add to them; we will not silently redefine one.

Event types

grant_new · grant_renewal · sbir_phase2 · trial_registered · trial_phase_advance · trial_recruiting · trial_completed · patent_granted · patent_assignment · raise_formd · fda_designation · entity_formed

Stages, earliest to latest

Grant-only · Formed · Preclinical · IND-enabling · Phase I · Phase II · Phase III · Filed/Approved

Therapeutic areas

oncology · neuro · immunology · rare · other

Errors

What you can do with it

The underlying federal filings are public domain and yours to do anything with. The resolved crosswalk, the momentum and runway scores, and the generated briefs are the thing you are licensing, and they are for your firm’s internal use. Redistributing them, or building a competing dataset out of them, needs a different agreement. The full terms, and our disclosures, which apply to the API exactly as they apply to the app.