Your AI team is building on Cognite. Your AIM workflows run on structured asset records. But for jackets, subsea, and pipelines, the condition intelligence still lives in PDFs locked inside inspection contractor file shares. Hullproof gives you JSON. Confidence-scored, frame-referenced, audit-ready — slotted directly into your existing data fabric.
Every link below is a live endpoint. Open in a new tab, save the payload, walk it through your data team. No demo signup, no gated trial.
/api/passports/demo-offshore-jacket-001/api/passports/demo-offshore-jacket-001/cdf/api/passports/demo-offshore-jacket-001/osdu/api/passports/demo-offshore-jacket-001/rdf/api/passports/demo-offshore-jacket-001/pdf/api/mcp/tools/platform/vqa-vlaHullproof emits a CoatingPassport JSON for every inspection. The /api/passports/{id}/cdf endpoint returns CDF Asset + Events JSON your team posts directly to your Cognite instance. No custom transformations, no per-asset bespoke ingest. Atlas AI agents consume the same Asset hierarchy you already use.
NV-Tesseract handles time-series anomaly detection (wells, heat exchangers, gauges). Hullproof handles footage-based structured condition data (jackets, subsea, pipelines, FPSO hulls). Same Cognite Atlas AI consumption layer, different signal channels. See /resources/nv-tesseract-coating-passport-complement for the integration pattern.
Yes — POST /api/passports/{id}/qa runs Visual Question Answering against the passport with Claude Haiku 4.5. Returns a structured answer with confidence + citations to specific finding_ids + source_frames. grounded_in_passport flag enforces no-hallucination. Live now.
Hullproof has a VLAProvider interface (src/lib/vlm/provider.ts) co-developed at platform-tier with operator engineering + robotics teams. The pattern: Hullproof finding → VLA orchestration → Spot / ANYmal / ROV / AUV action. See /platform/vqa-vla and /platform/robotics for the architecture.
Offshore platform jackets (NORSOK N-005, DNV-RP-C203, PSA Norway §47), subsea structures (NORSOK U-009, DNV-RP-B401, API 17A), pipelines (DNV-ST-F101, API 1160), wind monopiles, plus maritime hulls / ballast tanks / cargo holds for FPSO and shuttle-tanker fleets. Adding an asset class is a directory in our codebase — every new vertical lands on the same canonical passport.
Every finding carries explicit confidence, supporting frame count, image quality score, AI model version, and source frame references. Every passport tags its EU AI Act class (high-risk-AI-system by default for compliance use). Review workflows are tracked in append-only history. The structure IS the audit trail.
Hullproof is footage-agnostic. ROV, work-class ROV, drone, satellite, fixed monitoring, handheld — anything a VLM can read condition from. No hardware lock-in, no vendor coupling. Use your existing IMR contractors.
Two tiers fit operators: Enterprise API ($80K-300K/year, fleet-wide JSON access + custom integrations) and Platform ($500K-2M/year, OSDU/CDF connectors, VLA pipeline, multi-vertical). Pricing scales with assets covered, not API calls.
Yes. Single asset, single inspection cycle, real footage from your team, full passport + Cognite CDF integration. From contract to live ingest is short. After that, scale to fleet.
One asset. One inspection cycle. Real footage. Live Cognite CDF integration. After it works, scale to fleet.