{"resourceId":"ub-jaggaer-jai-procurement-assistant-pilot-evaluation-202610","versions":[{"version":"external-0db0e6807f650c26e05c1d798a4538098c8162973ef66a6dcd2a8e79e0b55ec0","resource":{"id":"ub-jaggaer-jai-procurement-assistant-pilot-evaluation-202610","title":"University at Buffalo's pilot of an AI procurement assistant inside its eProcurement system ends this month, with user ratings deciding its future","organization":"University at Buffalo (SUNY)","sector":"Higher education procurement","geography":"New York, United States","publishedAt":"Undated web page; pilot stated to run until the end of October 2026","publicationDate":null,"eventDate":null,"sourceName":"University at Buffalo Administrative Services Gateway","sourceLabel":"Official institutional guidance","sourceUrl":"https://www.buffalo.edu/administrative-services/managing-procurement/shopblue-system/ai-assistant-for-shopblue.html","evidenceClass":"standards-guidance","outcomeClass":"emerging","topics":["knowledge-work","governance-procurement","developers-agents"],"finding":"UB is piloting JAI, an AI assistant built into its JAGGAER-based ShopBlue procurement system, until the end of October 2026. It answers from UB-approved policies and training, explains approval workflows, and can show a user's recent purchase orders and receivers. Users rate every response, and UB says the evaluation will decide broader availability. The page states JAI complements rather than replaces existing support channels.","sledRelevance":"Embedded vendor AI in procurement suites is arriving on many campuses; UB's pilot ends now, so its go or no-go decision is a near-term signal. Results are not yet public, limiting what can be concluded.","evidence":"Official procurement page describing scope and evaluation process; no results yet.","architectureImplications":"Assistant embedded in the procurement suite with read access to a user's purchase orders and receivers.","governanceImplications":"Time-boxed pilot with per-response ratings feeding a decision on availability.","securityPrivacyImplications":"Read access to purchasing records means answers must respect each user's permissions; the page does not detail controls.","caveats":"No results, criteria or costs published; page undated; vendor co-developed the pilot.","streamIds":["campus-operations"],"roles":{"sales":"Sales framing: UB's procurement AI pilot ends this month, and its future depends on user ratings. Interpretation: procurement and finance leaders evaluating embedded vendor AI may want a structured way to decide on continuation beyond thumbs-up counts. A bounded engagement is a pilot evaluation framework for embedded procurement AI. Do not claim efficiency gains; UB has published no results.","engineering":"Engineering framing: JAI is embedded in the procurement suite, grounded in UB content and able to show a user's recent purchase orders. Interpretation: pre-sales should check whether the assistant enforces the same record permissions as the suite, how grounding sources are curated and updated, and whether rating data can be exported. The page names no controls, so these are open checks.","delivery":"Delivery framing: UB set a fixed end date and a rating loop for its pilot. Interpretation: delivery teams running similar pilots should define accuracy and deflection targets before launch, sample rated answers for review, and log which policy pages answers cite. Acceptance should combine ratings with expert review, because ratings alone do not establish correctness."},"roleDetails":{"sales":{"customerProblem":"Campuses must decide whether embedded procurement AI is worth continuing after a short pilot.","stakeholders":["Procurement director","Controller","Procurement systems manager","Chief information officer"],"discoveryQuestions":["What criteria decide continuation?","Does the assistant reduce help requests?","Is there an added license cost?"],"valueHypothesis":"Interpretation: a clear evaluation could support a renewal decision; no value is shown yet.","potentialEngagement":"A bounded evaluation framework for an embedded procurement assistant pilot.","unsupportedClaims":["Time savings","Fewer procurement help tickets","Accuracy of answers"]},"engineering":{"fit":"Fits campuses on procurement suites that offer embedded AI assistants.","architectureIntegration":"Assistant within the eProcurement suite grounded in institutional content with read access to purchase records.","prerequisites":["Curated policy content","Role-based access in the suite","Rating data export"],"constraints":["Controls undocumented","Vendor-hosted feature"],"security":"Assistant must respect suite permissions when showing purchase orders and receivers.","proofOfValue":["Test that users see only their own records","Review a sample of rated answers against policy"]},"delivery":{"work":"Run a time-boxed evaluation of an embedded procurement assistant.","dependencies":["Vendor cooperation","Procurement content owners","User pilot group"],"ownership":"Procurement systems and operations team.","skillsAdoption":"Users trained to rate answers and suggest questions.","governanceCheckpoints":["Evaluation criteria set before pilot","Expert review of rated answers","Go or no-go decision documented"],"acceptanceCriteria":["Proposed: at least 90% of sampled answers match policy","Proposed: measurable drop in procurement help requests"],"risks":["Ratings mistaken for accuracy","Stale policy content","Unclear cost after pilot"]}},"retrievedAt":"2026-10-29T21:50:11-05:00","enrichedAt":"2026-10-29T21:53:56-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Positioned as complementing procurement staff support, not replacing it.","procurementImplications":"Vendor feature in an existing contract; cost of continued use not stated.","operatingModelImplications":"Procurement systems staff own the evaluation and its decision.","sourceVerification":{"openedUrl":"https://www.buffalo.edu/administrative-services/managing-procurement/shopblue-system/ai-assistant-for-shopblue.html","referenceExcerpt":"JAI is currently in a pilot phase until the end of October, 2026.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}