How-toERP selection and AIProcurement / RFP requirements

ERP RFP Template with AI Requirements

Question

ERP RFP template with AI requirements

Also searched as

ERP RFP requirements template AI section
AI requirements list for ERP selection
ERP RFP checklist AI capabilities
sample AI requirements ERP request for proposal

Short answer

Add six AI requirement blocks to your ERP RFP: data access and grounding, security and roles, deployment and data residency, model governance, user experience and explainability, and commercial terms. Each block below has requirement wording you can paste and a way to mark it mandatory, weighted or informational.

Applies to: ERP replacement, ERP upgrade and AI add-on RFPs; works with any requirement spreadsheet format (Excel, procurement portal, RFP tool)

How to use this template

1Copy your existing functional requirement spreadsheet and add a new tab named AI. Keep the same columns: ID, requirement, priority (M mandatory, H high, S standard, I informational), vendor response code, comments.
2Paste the requirement blocks below into the tab and delete rows that do not apply, for example the air-gap row if you run in a public cloud.
3Mark deployment and data residency rows as M so non-conforming bids are screened before scoring.
4Add a column 'Evidence' and require a document reference, architecture diagram or live test for every M and H row, not a bare yes.
5Add a response code 'Roadmap' with a required delivery date and make it count as zero in scoring unless contract terms commit to it.
6Attach your scenario list and pricing template as appendices.
7Send the draft to security, legal and an ERP power user for review before issue.

Block 1 and 2: data access and security

AI-01: The solution shall answer questions using current data read from the ERP at query time, and shall state the source tables, API or report used for each answer. AI-02: The solution shall show the generated query or retrieval path to the user on request. AI-03: The solution shall support read-only access by default; any create, update or delete action shall require explicit user approval and be written to an audit log.

SEC-01: The solution shall enforce the requesting user's existing ERP roles and data-level restrictions, and shall not use a shared super-user credential to answer user questions. SEC-02: The solution shall support single sign-on with the customer's identity provider. SEC-03: The solution shall log every prompt, retrieved record set and response with user and timestamp, with retention configurable by the customer.

These rows catch the commonest failure: an assistant that connects with one broad service account and can therefore show a clerk the payroll or margin data the ERP would have hidden.

Block 3 and 4: deployment and model governance

DEP-01: State all deployment options (customer on-premises, customer private cloud, vendor cloud, air-gapped). DEP-02: State for each component (model inference, embeddings, vector store, logs) where data is processed and stored. DEP-03: The customer's data shall not be used to train or improve shared models. DEP-04: The solution shall operate without outbound internet access if required (mark mandatory only if true for you).

MOD-01: Identify every model used, its licence and its version, and the process for upgrading or replacing it. MOD-02: Support open-weight models served on customer-controlled GPUs where requested. MOD-03: Describe how answer quality is evaluated and regression-tested after a model or prompt change. MOD-04: Describe known limitations and the controls for incorrect or fabricated answers.

The /tools/on-prem-ai-security-checklist and /tools/ai-model-license-compliance-checklist tools list further items worth turning into rows for regulated environments.

Block 5 and 6: user experience and commercial

UX-01: Answers shall cite the records or documents used so users can verify them. UX-02: The solution shall allow users to give feedback on an answer and administrators to review it. UX-03: Users shall be able to export results to Excel or CSV. UX-04: Describe supported languages for questions and answers.

COM-01: Provide pricing by the stated unit (user, query, token, GPU) with a worked example at expected volume and at three times that volume. COM-02: State what is included in subscription and what is separately chargeable. COM-03: Provide data and configuration export on termination, in documented formats. COM-04: State the support model and response times for AI-specific incidents such as wrong answers or model outages.

Weighting the AI tab

Do not let the AI tab swamp the ERP fit. A common split for a replacement project is 80 to 90 percent core ERP and 10 to 20 percent AI, because AI layers can be changed later more cheaply than the ERP of record. For an AI-only procurement against an existing ERP, flip the weighting but keep security and deployment as pass/fail gates.

Pair this template with the /tools/erp-rfp-completeness-checklist and /tools/erp-selection-scorecard tools to check coverage and to turn responses into a ranked comparison.

Common pitfalls

!Accepting 'yes' in the response column without evidence for mandatory AI rows.
!Counting roadmap items as if they were delivered features.
!Leaving role enforcement out, then finding the assistant bypasses ERP security.
!Making every AI row mandatory so that no bidder qualifies.
!Copying requirements for capabilities you will not use, which inflates price and bid effort.

How an ERP-grounded AI assistant handles this

A grounded assistant over your ERP can speed up the template: ask it which reports and inquiries users run most and where they export to Excel, and add requirement rows for those. For SyteLine shops, SyteRay-style tooling can inspect existing customisations and reports to show which integrations a new vendor must replicate. Verify its findings against the system before issuing the RFP.

See ERPray

Frequently asked questions

How many AI requirements should an ERP RFP have?

Around 25 to 50 rows is usually enough: a handful per block above, plus scenarios. More rows mostly generate boilerplate answers. Prioritise rows that test architecture and data handling, since those are hardest to change after signing.

Which AI requirements should be mandatory?

Deployment and data residency, role enforcement, audit logging, no training on your data, and read-only default. Everything else can be weighted. Mandatory rows should reflect real obligations of yours, not a wish list.

Is there a standard AI requirements list for ERP?

No single recognised standard exists for ERP AI. Use frameworks such as NIST AI RMF or ISO/IEC 42001 as reference for governance rows, and your own security policy for the rest, then adapt wording to your ERP.

Can the same template work for a SaaS ERP?

Yes, with changes. Keep the data, security and commercial blocks, and replace on-prem rows with tenant isolation, sub-processor lists and the vendor's model provider terms. Ask which AI features can be switched off per tenant.

CIO briefing

How to Write an RFP for ERP AI

Write the ERP AI RFP as two linked scopes: the ERP platform and the AI layer that reads or acts on its data. Define the business questions the AI must answer, where data may and may not travel, which ERP roles apply, and how you will test answers before you sign. Vague asks like 'must have AI' produce slideware, not comparable bids.

How-to

How to Score ERP AI Vendors in an RFP

Score in two passes: pass/fail gates first (deployment, data residency, security, mandatory functionality), then weighted criteria on a defined 0 to 4 scale with evidence required for every score. Fix the weights before bids arrive, score independently, then reconcile, and keep price as a separate, calculated factor.

CIO briefing

Does Your ERP Vendor's AI Copilot Increase Lock-In?

Native vendor copilots such as SAP Joule, Oracle Fusion AI agents and Microsoft Copilot in Dynamics 365 are usually bundled or low-cost to start, but they only see that vendor's native data model and typically bill on consumption, which raises both switching cost and long-run cloud spend. Use them for what they do well inside the vendor's own UI, and keep cross-system or custom-field questions on a vendor-neutral layer.

CIO briefing

ERP AI Vendor Due Diligence: What CIOs Should Actually Check

The single highest-signal question in ERP AI vendor due diligence is 'show me it answering a real question against our actual ERP data, live, not a demo dataset.' Beyond that, focus diligence on data grounding method, security and access model, exit/portability terms, and references from customers on your specific ERP platform - not on feature checklists, which most vendors can match on paper.

CIO briefing

ERP Upgrade or AI First? A CIO Decision Framework

In most cases, add a grounded AI layer on top of your current ERP first, because it proves value in weeks and shows exactly what an upgrade would need to fix. Reserve a full ERP replacement for cases where the platform itself is end of support, unsupported, or structurally blocking the business, not simply because it feels dated.

CIO briefing

How Much Does ERP AI Actually Cost?

A scoped pilot against one ERP module typically runs 15,000 to 50,000 USD; a single-department production deployment runs 50,000 to 150,000; a multi-module enterprise rollout runs 150,000 to 500,000 or more, plus ongoing hosting and usage costs. Native vendor copilots are usually priced per active user per month on top of existing licensing, not included free.

AI for ERP