A step-by-step evaluation process that takes you from requirements gathering through vendor selection and into implementation planning. Written for procurement teams that have never bought an AI agent before — and those that have been burned before.
Most procurement AI evaluations fail because they skip steps or run them in the wrong order. This framework gives you a repeatable, rigorous process that works regardless of which category of AI agent you are evaluating.
Before looking at any vendor, define what success looks like. The guide provides a requirements workshop template that takes your team through current-state process mapping, pain point prioritisation, measurable success criteria, and stakeholder alignment. The most common reason AI evaluations go wrong is that procurement teams define requirements after they have already fallen in love with a demo. This stage forces rigour before vendor contact.
Outputs include: prioritised requirements list, current-state process flow, success metrics with baseline measurements, and a stakeholder alignment document for IT and Finance.
✓ Requirements Workshop Template IncludedBuilding a structured longlist using our category taxonomy and vendor profiles. The guide includes a 25-question demo script specifically for procurement AI vendors — questions that reveal real integration depth, AI model accuracy, support responsiveness and roadmap credibility that polished vendor demos are designed to obscure. Know what to ask before you sit down with any sales team.
Key demo script topics: ERP integration architecture deep dive, UNSPSC classification accuracy test cases, SLA and support commitments, AI model training data provenance, data security and residency.
✓ 25-Question Demo Script IncludedHow to design a 30-day POC that tests the specific capabilities that matter to your use case using your real procurement data. The guide covers POC scope design, test data selection, success metric definition, and evaluation rubric. Most procurement teams run POCs without defined success criteria — which means vendors control the narrative of whether the POC succeeded.
Outputs include: POC scope document template, test data preparation checklist, success scorecard, and evaluation team sign-off process.
✓ POC Framework Template IncludedBuilding the ROI case for CFO approval and negotiating AI SaaS contracts that protect procurement's interests. The guide covers ROI calculation methodology (savings, efficiency, risk reduction), TCO modelling, and the specific contract terms to push for in AI SaaS agreements: data portability, AI model transparency obligations, SLA penalties, and exit provisions.
Includes a business case template with worked numerical examples and a 12-point AI SaaS contract negotiation checklist.
✓ Business Case Template + Contract ChecklistA 40-item implementation checklist covering ERP integration configuration, master data preparation, change management planning, user training approach, and go-live readiness criteria. This stage covers the work that happens after contract signature — and it is where most implementations fall behind schedule and over budget. The checklist keeps every stakeholder accountable.
Outputs include: implementation workback schedule template, data migration readiness checklist, change management communication plan, and go-live decision framework.
✓ 40-Item Implementation Checklist IncludedThe most common evaluation gaps that lead to poor procurement AI investments — this framework is specifically designed to fill them.
"The POC design template in Stage 3 changed how we ran our evaluation. We went in with defined success criteria and a structured test data set. When the vendor's AI missed classification accuracy on our specific spend categories, we had the evidence to walk away — something we could not have done without defined benchmarks."
"The contract negotiation checklist is something I wish I had had on our last software purchase. The AI model governance clauses and data portability provisions saved us from terms that would have locked us in completely. We negotiated a data export right that the vendor initially refused — and now we have it in writing."
"I lead procurement transformation engagements and this framework has become a client deliverable on every AI tool selection project. The requirements workshop template takes what normally takes two weeks down to two days. The ROI model gives Finance the structure they need to approve the investment."