Asking bigger questions of your Maximo data, and AI Tester in early access

An answer drawn from the first screenful of Maximo records looks exactly like one drawn from all of them. Most of our last few months went into closing that gap. AI Tester, the other thing we are building, is now onboarding teams.

Interpretos: widening how much Maximo data one question can reach

A reliability engineer wants the failure rate of a family of assets set against PM compliance over the last eighteen months, before Thursday's review. A planner needs every open work order blocked on parts for over fourteen days, by craft, before tomorrow's plant meeting. A maintenance head wants the ten assets draining labour hours this quarter, before the budget conversation. None is exotic. Each needs years of records to answer, and the standard screens show a page at a time.

A question can now draw on far more than the on-screen sample of records, and reach Maximo modules the standard screens do not show. It does that through Maximo's own APIs, read-only, under the asking user's existing Maximo permissions, so nobody sees anything their login does not already allow. The fourteen-day parts-blocked list above runs across the full backlog, and the eighteen-month failure-rate question across the full history. The answer still arrives the way it must: verifiable against Maximo, showing its working and citing the records it used, wherever the record carries a source.

No AI model can hold a whole Maximo estate at once. Summarising a sample is worse than it looks: it reads like a finding about the whole estate and is not one. So every answer has to check out against the live system.

AI Tester: plain-English checks back as an evidence pack, now in early access

A test manager has a list of checks nobody has the hours to get to. She needs to prove the new approval workflow routes a purchase order over the limit to a second approver, on three sites, before the release goes in at the end of the month. She knows exactly what needs proving. Nobody is free to prove it and write down what happened.

The hard part for an AI is following her checks through a Maximo screen it has never seen, deciding whether what happened counts as a pass, and leaving evidence a colleague can check. A run that reports a pass and cannot show why is worth nothing on a release call.

AI Tester is in early access now. You describe the check in plain English, and what comes back is an evidence pack a colleague who was not watching can check: what was tried, what happened, and whether it passed.

Learn more about AI Tester and early access, or what we are building for questions over Maximo data. Bring the question your reporting has never answered, or the check nobody has the hours to run, and we will tell you which of the two fits.