Skip to main content
PRODUCT · UPSTREAM

Search the global patent landscape. Decide in 8 minutes.

IP Radar cross-references 130M patents across Espacenet, USPTO, EPO PATSTAT, and Google Patents BigQuery — then reads claim 1 of each shortlisted patent and returns FTO verdicts, white-space maps, and a deposit-ready intelligence report.

Start a landscape See it for your role
Espacenet OPSUSPTO PEDSEPO PATSTATGoogle Patents BigQuery
What it does

The landscape engine, in six capabilities.

How it works

Six steps from brief to deposit-ready ZIP.

  1. 01

    Brief

    Natural-language research question — drug class, target, formulation, indication. Up to 8000 chars.

  2. 02

    Search plan

    Gemini emits a ResearchPlan: an ordered list of typed subtasks with explicit criteria.

  3. 03

    BigQuery

    Each subtask compiles to a Google Patents BigQuery query. Reranker filters at 0.3 relevance.

  4. 04

    Claim read

    Per-patent Gemini call: claim 1 parsed into ClaimExtraction (limitations, scope, assignee, family).

  5. 05

    FTO verdict

    All-elements rule applied per claim against your product description. Red = every element matches.

  6. 06

    ZIP export

    Full IntelligenceReport JSON + per-drug briefs + raw subtask results — auditable, citable, foreign-counsel-ready.

Technical depth

The actual API. No marketing wrapper.

One POST kicks off the pipeline. Subtask results stream back to Firestore, durable across Cloud Run instance restarts.

Request
research/start
POST /api/v1/ip-radar/research/start
Content-Type: application/json

{
  "brief": "Oral formulations of vonoprazan free base on
            mesoporous silica SBA-15, 20 mg dose, with HPMC
            coating — landscape + FTO for EU/US markets.",
  "max_patents_per_subtask": 60,
  "enable_claim_reader": true,
  "proposed_product": "20 mg vonoprazan tablet, SBA-15
            adsorbent, HPMC enteric coat, dissolution >= 80%
            at pH 6.8 in 30 min",
  "pricing_tier": "unlimited"
}
SubtaskResult
claim_extraction
// SubtaskResult — one step of the agentic pipeline
{
  "subtask_id": "st_3_substance_patents",
  "type": "substance_patents",
  "title": "Vonoprazan — substance patents 2014–2026",
  "patents": [
    {
      "patent_id": "US-9908899-B2",
      "assignee": "Takeda Pharmaceutical",
      "filing_year": 2015,
      "claim_extraction": {
        "claim_1_text": "A pharmaceutical composition...",
        "elements": [
          { "text": "vonoprazan or salt thereof",
            "user_has_it": "yes" },
          { "text": "an enteric polymer coating",
            "user_has_it": "yes" },
          { "text": "dissolution >= 70% at pH 1.2",
            "user_has_it": "ambiguous",
            "reason": "User spec gives pH 6.8 not 1.2." }
        ]
      },
      "fto_verdict": "yellow"
    }
  ],
  "raw_candidates": 1247,
  "elapsed_ms": 184320
}
Where it fits

Upstream of every claim you draft.

IP Radar feeds the Claim Interpretation Lattice with a curated landscape; the lattice then hands optimised claims to Doctrinal Hardening for screening.

DOWNSTREAM

Claim Interpretation Lattice

10⁵–10⁸ readings of every claim across six axes. Consumes the IP Radar landscape; emits the optimal wording.

DOCTRINE

Doctrinal Hardening

69 documented failure modes neutralised by construction — G 2/21, Amgen §112, In re Cellect.

Stop guessing the landscape.

One brief. Eight minutes. A deposit-ready intelligence report.

See pricing Built for your role