Kav AI Platform · KAP v3.2
Active Physical Intelligence™ for asset integrity.
The first platform to close the loop between what sensors see, what process data says, and what engineering codes require — delivering risk decisions in hours, not weeks.
The integrity gap
Integrity data exists. Decisions don't.
Inspection, SCADA and engineering records live in separate silos. The people responsible for facility safety spend days manually reconciling them — while damage mechanisms keep running.
<10%
of captured inspection imagery is reviewed by a qualified engineer under current workflows.
5–10d
to move from inspection capture to actionable integrity decision across mid-sized facilities.
$2M
a day — the cost of an unplanned shutdown at a mid-sized refinery, at the top of the range.
What Kav AI is
The integrity intelligence layer above existing systems.
KAP ingests visual inspection data, reads operational data from SCADA and process historians, and reasons across both in a persistent 3D model of the facility — continuously updating risk and recommending action.
What Kav AI is not
Observe · reason · recommend
See, reason, recommend — in one system.
01
See
Multi-modal capture, normalised into a single photorealistic 3D model.
- — RGB · Thermal · OGI
- — Drone & confined-space fleets
- — Autonomous robot patrols (KRSI)
- — Fixed sensors & SCADA (read-only)
- — P&ID, line-list, materials (CAD)
02
Reason
Multimodal AI correlates visual, thermal and process data against the integrity domain model.
- — API 584 IOW classification
- — API 571 damage mechanisms
- — API 581 risk scoring + P90 life
- — Stage 3.5 consistency gate
- — Out-of-distribution detector
03
Recommend
Decisions surface in minutes via a natural-language interface — every action engineer-confirmed.
- — Prioritised inspection plans
- — CMMS work order packets
- — Chat over the 3D model
- — HITL sign-off required
- — Full audit trail, timestamped
Integrity analytical chain
IOW → DMR → risk → inspection plan, continuously.
Normalise
Multi-source data QC'd with ±500 ms timestamp alignment.
IOW check
SCADA scored against API 584 limits by duration × intensity.
DMR
API 571 damage mechanism mapping, enriched with CAD context.
Cross-check
Campaign, patrol and SCADA evidence reconciled at the consistency gate.
Validate
Physical validation against UT, thermal, OGI and robot-sourced NDT.
Act
API 581 risk + P90 life → prioritised inspection plan & CMMS work order.
Result Triage-to-work-order cycle reduced from 5–10 days to under 4 hours.
Competitive position
Four layers. One system. No one else has all four.
| Layer | RBI incumbents orKsoft · Meridium |
Data fusion Cognite · Hexagon |
Autonomous capture Percepto · Flyability |
Kav AI |
|---|---|---|---|---|
| Visual inspection (multi-modal) | — | partial | ✓ | ✓ |
| Photorealistic 3D (from drone imagery) | — | needs LiDAR / CAD | — | ✓ 3DGS |
| SCADA/IOW context (API 584) | ✓ | ✓ | — | ✓ |
| Damage mechanism reasoning (API 571) | ✓ | — | — | ✓ |
| Deployment time | 6–18 mo | 6–18 mo | weeks | 90-day pilot |
Timing
Four technologies crossed the line in 2024.
KAP was not possible at scale three years ago. It is now.
01
3D Gaussian Splatting
Navigable, photo-quality facility models built from standard drone imagery — at 1/100 the cost of LiDAR.
02
Multimodal AI reasoning
Large models crossed the threshold for correlating visual, thermal and sensor data in a single analysis.
03
Vendor-neutral OPC UA
Read-only SCADA access across Emerson, Siemens, AVEVA, Ignition — without vendor lock-in.
04
Plug-and-play AI tools
New integration standards let AI call specialised detection models as tools — no custom pipeline work.
The value case
What a mid-size refinery unlocks.
Recurring savings fund adoption. Avoided major events create the asymmetric upside.
Hard savings
$3–8M /yr
Inspection optimisation, LDAR labor efficiency, turnaround scope control.
Operational value
$5–20M /yr
Downtime reduction, faster root-cause analysis, faster decisions.
Risk avoided
$50–150M+
Per major event avoided — business interruption, repairs, insurance impact.
Payback target
12–18 months, validated through a 90-day site pilot.
Solomon Associates 2023
OI.Expert × Kav AI
Six cost categories. Real evidence per claim.
Mid-size refinery, 200–250 kBPD, North America. Each row carries an evidence tier — confirmed, derived, or estimate.
01 · Routine inspection
Estimate$5–8M /yr
NDE, corrosion monitoring, DMR, IOW, RBI upkeep. Solomon RAM: facilities >2% PRV are over-spending.
→ KAP target: $1–2M/yr
02 · Major turnaround
Estimate$150–250M /TA
Every 6–8 yrs. McKinsey: poor pre-TA inspection drives 15–25% overruns and 5–15 extra days offline.
→ KAP target: $1–5M/yr annualised
03 · Unplanned downtime
Derived$3–15M /yr
US DOE: 92% of mechanical shutdowns are unplanned. McKinsey: LOC events drove ~75% margin spike.
→ KAP target: $3–15M/yr
04 · Insurance premium
Derived$5–15M /yr
Repeat LOC / PSE incidents drive material premium escalation. CSB orders impose multi-year obligations.
→ KAP audit trail + SOC 2 supports renewal
05 · LDAR & emissions
Derived$1–15M+ /event
EPA / TCEQ fines plus remediation per major flaring or LOC event. Carbon penalties rising under US/CA frameworks.
→ KAP target: $0.5–1M/yr LDAR saving
06 · RCA latency
Confirmed$1.2–3M /day
Valero St. Charles: every day of TA overrun. McKinsey: RCA without continuous history takes 5–10 days longer.
→ KAP target: $2–10M/yr
Savings model
$5–15M/yr recurring. $20M+ with performance.
For each pain point: what KAP delivers, the reduction, and the evidence underwriting it.
| Cost category | Industry baseline | KAP saving | Reduction | Evidence |
|---|---|---|---|---|
| Routine inspection optimisation | $5–8M/yr | $1–2M/yr | 20–25% | Confirmed |
| Turnaround scope & overrun | $150–250M/TA | $1–5M/yr | 5–10% annualised | Derived |
| Unplanned downtime prevention | $3–15M/yr | $3–15M/yr | Significant | Derived |
| LDAR optimisation (OGI stack) | $0.5–1M+/yr | $0.5–1M/yr | Material | Derived |
| Insurance premium reduction | $5–15M/yr | Directional | At renewal | Estimate |
| Environmental fines avoidance | $1–15M+/event | Fewer Tier-1 | 30%+ fewer | Derived |
| RCA & decision latency | $1.2–3M/day | $2–10M/yr | 5–10 days | Confirmed |
Total annual value target
$5–15M recurring · up to $20M+ with performance
Investment / payback
~$1.5M · 12–18 months
Year 0–5 investor trajectory
Value compounds as the facility model deepens.
Each campaign ingested → facility model improves → savings compound. Year 1 anchored on confirmed AOC + LDAR; Year 5 illustrative.
−$1.5M
$4.5–13M
$7–18M
$10–28M
$12–32M
$13–36M
Y0
Pilot
Y1
Establish
Y2
Deepen
Y3
Optimise
Y4
Compound
Y5
Steady state
Year 5 cumulative
$45M – $126M
Payback
12–18 months
Confirmed anchor
AOC RBMI · $6.37M/yr · $63M / 10 yrs
90-day proof-of-concept
Proof before commitment.
Not a demonstration. It runs against the client's real inspection data, real facility, real operational questions.
Weeks 1–2
Onboarding
Data ingestion, 3D model generation, OPC UA connector, OOD calibration.
Weeks 3–4
Baseline
Initial detections cross-validated against known historical defects.
Weeks 5–8
Active use
NL queries on real integrity questions. IOW/DMR chain activated.
Weeks 9–10
Validation
Findings cross-validated. False positive/negative rates measured.
Weeks 11–12
Evaluation
Structured debrief. Calibration curve delivered. Proceed / extend / conclude.
Pilot success criteria
Success, measured in writing.
| ID | Criterion | Threshold | Measurement |
|---|---|---|---|
| SC-1 | Anomaly detection recall | ≥ 80% | Cross-validation against existing inspection records. |
| SC-2 | False positive rate | ≤ 20% | Pilot lead review of all flagged items. |
| SC-3 | Query response quality | ≥ 70% useful | Structured query log at Weeks 6 and 12. |
| SC-4 | Time to first finding | < 48 hours | From completed data ingestion. |
| SC-5 | AI response latency (P95) | ≤ 5 seconds | Automated latency logging. |
| SC-6 | SCADA correlation | ≥ 1 confirmed | IOW exceedance ↔ visual anomaly at same asset. |
From zero to live in under a year
M0 → M3, shipped on real data.
M0
Jul 2025
Platform foundation
3D viewer, auth, image gallery, operator dashboard — validated on first real inspection dataset (RGB).
M1
Dec 2025
AI foundation
Multimodal AI pipeline, NL interface, machine vision engine — tested on thermal + gas sensor data.
M2
Mar 2026
App MVP
Production-ready 3D viewer + AI chat unified in a single operator interface. Live alpha.
M3
Jun 2026
AI Q2 delivery
Contextual data chat, sensor-native analysis, chat-with-3D-map, interactive overlays, automated reports.
Third inspection campaign planned at an oil refinery — OGI imagery, calibrated thermal, expanded gas suite, first repeat-patrol comparison.
Compounding advantage
Every campaign compounds the moat.
Kav AI's advantage is not static. It widens with each inspection campaign, each patrol cycle, and each operator-confirmed finding — at each facility.
A customer 10 campaigns deep has a detection model tuned to their specific equipment, corrosion patterns, route coverage and operating conditions. A new entrant faces the same cold-start problem we've already solved.
With OI.Expert
AI plus human engineering validation.
OI.Expert provides the corrosion and mechanical integrity expertise that ensures Kav AI's outputs are validated, acted upon, and embedded into the client's inspection programme.
×
OI.Expert
Damage Mechanism Reviews (DMRs)
For asset classes covered by the pilot scope.
IOW recommendations & deviation response
Integrity Operating Window engineering support.
HITL validation panel
Integrity Engineers confirm Stage 3.5 / 4 findings on Critical mechanisms (HIC, NH₄Cl, Creep).
Inspection programme integration
Embedding findings into RBI, IDMS & compliance (API 580 / 581 / 584 / 510 / 570).
Fitness-for-Service & root cause
Reactive engineering, FFS, RCA, materials & welding consultation.
Let's talk
Book a 90-day pilot.
Structured, time-boxed, with success criteria agreed before data ingestion begins. If we meet them, the pilot fee credits toward Year 1. If we don't, you walk — data deleted within 5 business days.
Get in touch
contactus@kavai.com
Web
www.kavai.com