ERP AI Data Source Inventory Worksheet
Before an AI assistant can answer questions over ERP data and documents, you need a plain list of what data exists, who owns it, how sensitive it is and whether AI may use it. This worksheet is that list, in a form you can edit and export to Excel for the data governance board.
| Data source | System type | Data owner (role) | Classification | Approx. records or documents | Refresh | AI use decision | Access method | Quality score (1-5) | Notes | |
|---|---|---|---|---|---|---|---|---|---|---|
Start with the sources you expect the first use case to need, then widen. Never connect a source until the owner, classification and AI use decision are filled in. Quality is a quick 1-5 judgement of how clean and current the data is.
Turn my inventory into a design
Optional: an expert maps your inventory to a retrieval and permissions approach.
Optional. The tool and every export work without this. No spam.
How to use the worksheet
Add one row per source: an ERP module or reporting view, a document library, a file share, an email archive. Name an owner by role, pick a classification, estimate size, and choose an AI use decision. Use Allowed with filters where only part of a source should be visible, such as hiding supplier pricing from non-buyers. Review needed means someone with authority must decide before any connection is built. Export and circulate to data owners for sign-off.
Classification and AI decisions
Align classifications with your existing scheme if you have one. For manufacturers, the two classes that most often block progress are export controlled technical data and personal data. Export controlled data may be restricted by who can access it and where it can be processed, so confirm requirements with your compliance officer before any AI system indexes it. Personal data brings privacy duties such as purpose limits and retention.
- Allowed: low sensitivity, clean enough, owner agrees
- Allowed with filters: needs row, column or role limits at query time
- Review needed: classification or hosting question unresolved
- Not allowed: out of scope or prohibited by policy or contract
Why quality and refresh matter
An assistant is only as good as the data behind it. A source scored 2 for quality, such as scattered documents with unclear versions, will generate confident wrong answers. Refresh frequency sets expectations: a nightly extract cannot answer what is on the dock right now. Record both so use case owners can choose sources that fit their questions, and so the project plan includes cleanup work where needed.
Common mistakes
The usual errors are listing only ERP tables and forgetting the documents people actually search, skipping the owner column, and treating the inventory as a one-time exercise. Review it each time a new use case is proposed, and keep access methods specific, such as a named read-only reporting view rather than direct database access.
Frequently Asked Questions
Do we need to inventory everything first?
No. Start with the sources needed by the first one or two use cases, plus any sensitive sources you want to explicitly exclude. Expand as new use cases are approved. A small accurate list beats a large unfinished one.
Who should own each row?
The business role accountable for the data, such as the procurement manager for purchase orders or the quality manager for procedures. IT is the custodian, not usually the owner, and should not make sensitivity decisions alone.
How do we handle export controlled data?
Mark it and send it to review. Requirements depend on regulations such as ITAR or EAR and your contracts, including who may access it and where it may be processed. Your compliance officer should decide before any indexing happens.
Can this feed our data catalog?
Yes. Export to Excel or CSV and import the key fields into your catalog or governance tool. The AI use decision column is often the new field catalogs lack.
If you want help turning this inventory into a retrieval and permissions design for on-prem AI, Netray can walk through it with your IT and data owners.
Related Tools
AI Model Card Template for ERP Use Cases
A one-page model card for each AI model or agent in your ERP landscape, covering intended use, data, limits, evaluation and ownership.
Compliance & ContractsAI Acceptable Use Policy Template for ERP Users
A fill-in acceptable use policy that tells ERP users what they may and may not paste into AI tools, and who approves exceptions.
AI Agents & AutomationAI Risk Register Worksheet for ERP Projects
An editable risk register with likelihood x impact scores, owners and mitigations, pre-filled with typical ERP AI risks.
Compliance & ContractsAI Vendor Security Questionnaire Template
A questionnaire covering data handling, model training use, access control, logging and incident response for AI vendors.
More AI Agents & Automation tools
See all- Business Process Automation ROI Calculator
- AI Agent Use Case Prioritizer
- Document Processing Automation Savings Calculator
- AP Invoice Automation ROI Calculator
- Order Entry Automation Savings Calculator
- Quote-to-Cash Cycle Time Calculator
- Manufacturing AI Readiness Assessment
- RPA vs AI Agent Selector
- ERP AI Maturity Assessment