Published on :
Views...
This digest covers the releases Xempla shipped across August and September 2026. The theme running through them is closure: making sure every finding, whether it comes from a verification report, an AI inspection or a fault, reaches an owner, an action and a verified outcome.
June and July connected the steps of the work. August and September were about making sure those connected steps actually finish.
The last digest set out four commitments for August. Here is where each one landed.
In UK healthcare, HTM 03-01 governs specialised ventilation. Systems serving critical areas such as operating suites are inspected quarterly and have their performance verified every year. Those verification reports contain findings that matter well beyond the plant room. In a clinical setting, ventilation is part of infection control.
The workflow now covers the whole journey. A verification report arrives and is read and analysed automatically. Any issues are surfaced in the governance layer and linked to a workflow in the DIIV cycle (Discover, Investigate, Implement, Verify), so every finding has an owner and a route to closure.
Why this matters: Verification has been failing in two different ways. The first is time. Engineers either didn't have time to review the reports, or reviewed them but couldn't act before the moment had passed. The second is visibility. Some findings are critical to compliance and commercial teams, yet those teams either had no access to them or lost them somewhere in the handoffs. Automating how reports are read addresses the first problem. Linking findings to the workflow they should trigger addresses the second.
What's changed so far: The workflow covers 150 locations in UK Health & Care, and around 70 verification reports have been processed automatically to date. For a service provider running it end to end, the result is higher engineering productivity, tighter governance and control, and better service delivery for the client.
Supervisors run quality checks and inspections every day. The long-standing problem has been what happens afterwards. An inspection records a problem, but nobody can be sure the record ever leads to anything in operations. Across August and September, we connected the full chain.
1. Inspection findings become tickets (August). When an AI inspection finds an issue at a location, the facility's own governance configuration decides whether the finding's priority justifies a ticket. If it does, the ticket is raised automatically and assigned to the supervisor and on-site team for corrective action. This is live across 5 facilities in India.
2. Quality governance from location to facility (September). Quality checks on each checklist now roll up to the location and then to the whole facility. This gives the service provider one view of whether the soft services regime is working and where the gaps are. It also surfaces the early signs of cost overrun, such as training gaps, upskilling needs and more deep cleaning than planned, so teams can act on them before they show up in the budget.
3. The automated business review. Most monthly business reviews are long documents that very few people read. This one is a two-page note generated automatically from what is happening in Xempla: checklist completion rates, quality check coverage, tickets raised, and recurring issues grouped by location and category. It can be scheduled monthly or weekly by the service provider, the owner-occupier or the managing agent.
Why this matters: A soft services provider can now show that its inspections lead to better outcomes, not just a record of what was wrong. For the client, it is the difference between seeing a quality score and seeing quality actually improve.
June and July brought the governance layer to executives. In August, we brought it to engineering managers, on both the mobile and web apps. The view covers six domains: lifecycle, reliability, maintenance, energy, operations and service delivery. . It shows which assets carry the most risk or the most opportunity to improve service delivery. It is live on every site where Xempla operates.
We also added human supervisory notes to risk and energy-saving assessments. The triage agent ranks every fault across several domains. Where the risk or opportunity is highest, a human reviewer adds a note. Planners can then see that a finding is both high-opportunity and human-reviewed, and prioritise it with confidence.
Why this matters: Agentic ranking works at a scale no team can match, and human judgment carries weight that no ranking can. Notes are added selectively, only where the stakes justify them, so when a note appears, planners know it deserves their attention.
Our design principle hasn't changed. When a work order is generated, everything a technician needs to do the job should already be there, without them having to ask. But technicians still have their own questions. They may be new to a site, or want to understand how an asset was configured or what happened to it before.
The conversational interface, released in August, is for those questions. It draws on the context available in Xempla, making information easier to find and helping close the skills gap for engineers who haven't built up years of knowledge about a particular site.
Xempla is now available to individual engineers, not only to organisations. Engineers sign up with a personal email address and receive 25 credits a month. They start with a complete asset inspection workflow that handles audio, video and other input types. It is equally suited to small and medium businesses.
It is currently in invite-only beta available globally, with wider access to follow.
It is part of a belief we have held since the start of Forever Forward: some tools should be accessible to everyone. We'll see how it goes.
We are continuing with the same thesis. The work under way includes:
We'll share what ships as it ships.
June and July connected the steps. August and September made sure the connected steps finish. A verification finding now reaches the person who has to act on it. An inspection result becomes a ticket, the ticket leads to a fix, and the fix appears in a two-page review a client will actually read. A fault ranking arrives with a human note where it matters most.
None of this adds new places to look. It makes sure that what the system finds goes somewhere.
Our thesis: A governance layer is only as good as how few systems a decision-maker has to move through to reach a decision, and how reliably that decision reaches an outcome. That is the System of Decisions, built one seam at a time.
© Xempla. All rights reserved.