Data Handling Commitment

The core commitment

We process your catalog. We do not keep it.

PartsCleanse AI is a diagnostic engine, not a data-hosting platform. When you upload a catalog CSV, the file is read into memory, analyzed for duplicate families, and used to generate report artifacts. The source file is then deleted. Your part descriptions, manufacturer numbers, vendor codes, and item master rows are never written to our database.

This is a deliberate architectural decision. Verdantis and Verusen operate as catalog enrichment platforms — they store and enrich your records as the core of their business model. PartsCleanse AI produces a governed diagnostic finding and destroys the source. You keep the output. We keep the metadata.

If your legal team is comparing vendors: The answer to “what do you retain?” is aggregate summary metrics, Open Findings, and report ownership. Not rows. Not descriptions. Not part numbers. We can put this in a countersigned DPA.

Retained vs. purged

Data element	Status	Catalog source purged?
Uploaded catalog CSV (source file)	Purged after report generation	Yes — immediate
Individual part descriptions and item numbers	Analyzed in memory, never written to database	Yes — in-memory only
Manufacturer part numbers, vendor codes, plant codes	Analyzed in memory only	Yes — in-memory only
Open Findings summary and duplicate group count	Retained, tied to authenticated user	N/A — retained
Diagnostic metadata (SKU count, runtime, exposure estimate)	Retained for dashboard	N/A — retained
Report ownership and access log	Retained for authentication and audit	N/A — retained
Quota usage	Retained to govern plan limits	N/A — retained

Processing lifecycle — six steps

From upload to purge: what happens inside the diagnostic engine.

1. Secure upload

Received over HTTPS into an isolated, session-specific run folder. No other user, session, or admin process has read access to that folder during or after processing.

2. In-memory analysis

The file is read into application memory for duplicate-family detection, description normalization, and confidence scoring. Raw catalog rows never enter the database.

3. Report generation

Five report artifacts are produced: browser Open Findings, Excel workbook, Word executive narrative, PDF summary, and clean deduplicated catalog CSV. Written to the run folder, not the database.

4. Source file purge

The uploaded source catalog file is deleted immediately after report generation completes. Industrial IQ does not keep, archive, or back up your original catalog rows.

5. Session-scoped downloads

Excel, Word, PDF, and clean CSV downloads are available only during the active generation session. After the session ends, only the authenticated browser Open Findings report remains accessible to the generating user.

6. Permanent retention only

Open Findings summary, SKU count, duplicate rate, exposure estimate, runtime, report ownership, quota usage, feedback, and audit metadata. No catalog rows. No item descriptions. No part numbers.

Quotable commitment — source purge

PartsCleanse AI purges the uploaded catalog CSV immediately after report generation. Raw catalog rows, part descriptions, and item numbers are never written to the database. Only aggregate diagnostic metrics are retained.

Competitive differentiator — data model

Verdantis and Verusen retain and host your catalog records as the core of their service model. PartsCleanse AI is a diagnostic engine: the source is processed, the governed finding is delivered, and the original file is destroyed. You own the output. We retain only the metadata required to govern the report and your account.

Data handling FAQ

Questions legal, procurement, and CISO teams ask before approving upload.

Who can access our uploaded catalog file?

Nobody except the PartsCleanse AI processing engine during the active analysis window. The file exists within an isolated session run folder with no admin, employee, or contractor read access. It is deleted before any human review of the diagnostic run is possible.

Can Industrial IQ use our catalog data for AI model training?

No. Your catalog data is never used for model training, benchmarking datasets, product improvement, or competitive analysis. The diagnostic engine is pre-trained. Your upload is consumed and destroyed within the same processing sequence.

What if our catalog contains sensitive part descriptions or export-controlled items?

The file is processed in memory and purged immediately after report generation. Descriptions, manufacturer part numbers, and vendor codes are analyzed for duplicate detection but never written to the database. The only retained data is aggregate metrics: SKU count, duplicate rate, and estimated exposure.

How does this compare to Verdantis and Verusen?

Verdantis and Verusen are catalog enrichment platforms — they retain, enrich, and host your catalog records as the core of their service model. PartsCleanse AI is a diagnostic engine: we process the data, purge the source, and deliver the finding. You own the output. We retain only the metadata required to govern the report and your account.

Is a Data Processing Agreement available?

Yes. A GDPR Article 28-compliant DPA template is available for immediate download without a sales call. Enterprise accounts requiring custom DPA terms or countersignature should contact support@ai2coe.com before uploading.

What jurisdiction does data processing occur in?

Processing occurs in the United States. For EU, UK, and GCC accounts with data localization requirements, contact support@ai2coe.com to discuss jurisdiction-specific arrangements before initiating the first diagnostic session.

Enterprise accounts — pre-upload legal review

DPA, custom data residency, and countersignature.

For accounts requiring a countersigned GDPR Article 28 Data Processing Agreement, a jurisdiction-specific data residency arrangement, or legal review before first upload: contact support@ai2coe.com with “DPA Request” in the subject line. A standard template is available for download without a call at /dpa.

Download DPA Template View Full Trust Center Security Overview