INDUSTRIAL IP PROTECTION · 2026

Your catalogue is protected. Yet AI already knows it.

AI doesn't need your catalogue. It infers it from thousands of public fragments — distributor pages, technical PDFs, forums, cross-reference tables. PartSentinel is the only audit that measures the gap between the catalogue you control and the catalogue the market is already consulting.

LIVE
NVX-22010 · audited 14:32 · sentinel score 31
SENTINEL SCORE · NVX-22010 (sample)
27/100
INDUSTRY MEDIAN · 42 · YOUR PEERS
Model A
19
✓ ID✗ XREF⚠ APP🔓 LEAK
Model B
41
✓ ID✓ XREF⚠ SPEC✗ APP
Model C
22
⚠ ID🔓 LEAK ×3✗ SPEC
Model D
26
✗ ID✗ XREF⚠ SPEC
THE CORE FINDING

You now have two catalogues.
One you control. One the market is already consulting.

AI doesn't need your catalogue. It reconstructs it from public fragments — and that reconstruction is what your buyers, technicians and distributors are starting to consult.

REAL CATALOGUE — YOURS

Internal, controlled, accurate

  • Official references, validated compatibilities
  • Protected by your systems and contracts
  • Updated through your internal processes
  • Visible only to authorised channels
INFERRED CATALOGUE — AI'S VERSION

External, reconstructed, uncontrolled

  • Invented equivalences, approximate compatibilities
  • Reassembled from public fragments without your consent
  • Your competitors may rank higher within it
  • This is the version your market is now starting to use

The gap between these two catalogues is now measurable, actionable — and directly correlated to your commercial visibility in conversational search environments.

THE STATE OF INDUSTRIAL IP EXPOSURE · 2026

Your IP is already public.
You just don't know where, or who has it.

73%
of catalogue references are exposed on unauthorized platforms or learned by at least one major LLM — based on 2.4M references audited across automotive, electrotechnical and aerospace verticals.
€15M
maximum AI Act fine for non-compliance with Article 53(1)(d) — deadline August 2, 2026. Without a structured exposure register, your rights are undefendable.
0
GEO and brand-monitoring tools that map IP exposure at the SKU level. They monitor your brand, not your parts.
HOW WE SEE WHAT YOU CAN'T

A digital mirror of your IP exposure.

We are a sentinel, not an optimisation tool. PartSentinel ingests your catalogue — BMEcat, ETIM, CSV, or PIM API — then crawls every major large language model and gray-market source as an external entity. Our AI thinks like a competitor or a regulator, not like your IT.

For each reference, we report which platforms list it without authorization, which models recall it, and — most importantly — which of your confidential cross-references have leaked into training data or scrape sources without your consent.

No content rewriting. No SEO games. No promises to “boost visibility.” A calibrated, auditable map of where your industrial IP lives in the wild — so you can finally protect what matters.

Read the methodology →
ANATOMY OF A SINGLE REFERENCE AUDIT
REFERENCENVX-22010
Exhaust silencer · Volvo FH
Model A
✓ accurate
Model B
⚠ partial
Model C
✗ hallucinated
Model D
🔓 leaked
Model E
⚠ partial
Model F
✗ obsolete
SENTINEL SCORE
31/100
THE SENTINEL SCORE

Like NPS, but for AI knowledge.

NPS — Net Promoter Score — is the canonical 0–100 customer-loyalty metric every CEO already tracks at a glance. The Sentinel Score gives you the same one-glance signal for how AI represents your catalogue: a single 0–100 number summarising accuracy across all audited models, calibrated against ground-truth data you control.

0255075100
YOUR SCORE
27/100
Coverage
31
Industry: 38
Accuracy
22
Industry: 41
Depth
14
Industry: 29
Freshness
38
Industry: 35
Comp. Exposure
61
Industry: 47
THE NVX-22010 (anonymised sample)

One reference. Six models. Six different stories.

An illustrative audit on a single exhaust silencer reference — brand and refs anonymised. Click each model to see its response, the cross-references it volunteered (some confidential), and the technical spec values it claimed.

REFERENCE · EXHAUST SILENCER (anonymised)
NVX-22010
HD truck Class A · Class B · Class C
REFERENCE SCORE
31/100
Raw response · prompt PT-AUTO-EX-V2.1
NVX-22010 is an exhaust silencer assembly used in HD Class A trucks. Material: aluminized steel housing with stainless internal baffling. Typical replacement interval is 350,000–500,000 km depending on duty cycle. No cross-references known.
prompt: PT-AUTO-EX-V2.1·temp: 0.0·tokens: 248 / 1024·cost: €0.014
Cross-reference diff · ground truth vs ai-stated
OE-A 21807576✓ MATCH
— (not stated)✗ HALLUCINATED
— (not stated)✗ HALLUCINATED
score · 19/100
Model A returned only 1 of 12 cross-references. Identification correct, but coverage is severely limited.
FOR INDUSTRIALS

Calibrated to your vertical.

Each industry has its own ground-truth conventions, cross-reference systems, and confidential data classes. PartSentinel's prompts and scoring weights are calibrated per vertical.

AUTOMOTIVE AFTERMARKET

Cross-reference confidentiality is the entire moat.

OEM catalogues, OE-numbers, fitment charts. Where leaks hurt most.

TYPICAL SCORE RANGE
35–55
For aftermarket →
ELECTROTECHNICAL & SUPPLY

Datasheets, pinouts, obsolescence chains.

ETIM classes. Models hallucinate spec ranges with high confidence.

TYPICAL SCORE RANGE
5–25
For electrotechnical →
AEROSPACE & DEFENSE

Tightly controlled catalogues, lower exposure.

ATA chapters, IPC. Coverage is sparse but accuracy where present is higher.

TYPICAL SCORE RANGE
65–85
For aerospace →
CHEMICALS & COATINGS

REACH, EINECS, MSDS — when AI invents, it invents a risk.

Hazard codes, formulations, regulatory recall. Hallucinations carry safety weight.

TYPICAL SCORE RANGE
30–60
For chemicals →
TRUSTED BY INDUSTRIAL DATA TEAMS · DACH · FRANCE · BENELUX
DIRECTOR OF PRODUCT DATA · TIER-1 EUROPEAN AFTERMARKET GROUP
We discovered three large language models knew our internal cross-reference codes — codes that exist nowhere on our public site. PartSentinel was the first tool that could even tell us which ones, and how recently.
EU AI ACT · ARTICLE 53(1)(d)
Deadline August 2, 2026. Generate your compliance pack in one click.
See AI Act compliance →
TIME-SENSITIVITY

The real question is not "do we need this audit?"

When do you want to find out?

NOW — WINDOW OPEN

Before your market reconfigures

The problem is still emerging. Gaps are measurable, inference paths identifiable, remediation accessible. The cost of early control is low compared to the stakes.

✓ Control possible
LATER — WINDOW CLOSED

After AI has reconfigured demand

Visibility loss has become structural. Competitors have captured category authority in AI answers. Commercial displacement is already underway.

✗ Damage embedded
TRY IT · 1 REFERENCE · 90 SECONDS · FREE

Type one reference. See what AI knows about it.

Six models queried live on your SKU. Sentinel Score, hallucinations and leaked cross-references — a full audit, free, no sales call.

No account required to start · Sign-up at the next stepor see a sample report →
  • 90 sec average
  • 6 / 339 models queried live
  • Free, no sales call