The second AI line in OpenText's VIM 26.2 notes is about the process rather than one document: "Get better transparency by letting AI answer any questions on invoices and generate process insights based on your query in natural language." On Tuesday we looked at the document questions, which read the invoice in the archive. This one reads your VIM process data.
That makes it a reporting feature with a friendlier front door. A controller who has never opened a VIM report can type "how long do non-PO invoices wait for approval in the UK?" and get a number back. The question sounds clear. It contains four choices nobody has agreed yet.
Every plain question hides a definition. "How long" from when: the email arrived, the scan, the document processing record was created, or the approval step started? "Wait" until when: approved, posted or paid? "Non-PO" by which flag? "UK" by company code or by supplier country? A dashboard forces someone to make those choices once, when it's built. A free-text question lets each person make them silently, every time they ask.
Write a short glossary before the feature goes live. These are the terms we'd define first:
Received. The choice to make: Channel arrival, or VIM document creation. Our suggested owner: VIM process owner.
Cycle time. The choice to make: Received to posted, or received to paid. Our suggested owner: Controller.
Exception. The choice to make: Any exception, or only those needing a person. Our suggested owner: AP lead.
Touchless. The choice to make: No human step at all, or no step after capture validation. Our suggested owner: VIM process owner.
Overdue. The choice to make: Past due date, or past your internal target. Our suggested owner: AP lead with treasury.
The owners are our view. The point is that each term has one owner and one answer, written where AP and finance can both find it.
Separate the channels first. Our earlier post on VIM reporting after go-live argued for splitting EDI, email, scan and e-invoice before trusting any blended score. That holds even more here, because a natural-language answer will blend by default unless the question says otherwise. If e-invoices flow touchless and scanned paper doesn't, one average hides both.
Test with questions you can already answer. Before rollout, take five numbers your current VIM reports or dashboards produce every month: backlog by exception type, average approval time, invoices posted per day. Ask the 26.2 feature the same questions in plain words. Where the answers match, you've learned how it reads your terms. Where they don't, you've found a definition to fix, in the glossary or in the question.
Ask OpenText two things. The blog doesn't say how the feature scopes its answers. Does it respect the SAP authorizations a user already has, so a clerk in one company code can't see another's backlog? And can it show the selection behind a number, so someone can check it? Both answers sit in the 26.2 release notes on OpenText My Support. Read them for your release, and confirm which 26.2 package fits your S/4HANA release, since OpenText's lifecycle page still lists 25.4 as the latest VIM release (read 31 October 2026; current maintenance for VIM for SAP Solutions ends 31 December 2030). If you're on 26.2 already, OpenText Support, KB0874931, describes a fix for several Fiori bugs reported with that release, including an AP Workplace error on Submit and Open (sign-in required). Apply it before you judge the new features.
Agree the glossary now, whatever release you're on. Your existing reports will get better from it.
Where does your landscape stand?
Take the free Assessment. It's about 10 questions, and you get a one-pager with suggested next steps and typical budget ranges, which aren't quotes.
