60% manual claims review eliminated for a UK healthtech firm in 12 weeks.
£40M-revenue UK healthtech firm. London strategy + Manila build. One business metric moved within the first 60 days.
Photo: Unsplash. Identifying client details anonymised per NDA.
Three claims operators reviewing the same data, four times.
The client — a UK healthtech firm processing supplemental health claims for ~280,000 covered lives across employer schemes — had a manual claims-review process that was eating margin. Every claim was touched three times: once for completeness, once for policy match, once for fraud-risk. 22 full-time claims handlers in a Reading-based ops centre. Average review time of 14 minutes per claim. 84,000 claims a year coming through the queue.
Leadership had evaluated AI consulting in the UK for nine months. Three big-four firms had pitched. Two had delivered "AI strategy" decks. None had reached production. The client's COO, frustrated, summarised it like this:
"Every consultancy told us what AI could do. None told us what they would actually ship — or when. We needed a builder with a strategy team, not a strategy team that subcontracts builders."
— COO, UK HealthTech firm (anonymised, ~280 employees)
Why a dual-geo AI transformation company won this engagement.
Two structural reasons. First, the engagement needed UK-side regulatory awareness (the client operates under FCA-adjacent insurance product governance, and the data set includes GP-issued clinical codes). A pure-play offshore AI shop would not have understood the regulatory contour. Second, the engagement needed real engineering throughput — claims handling AI is not a slide deck, it is a production system with audit trails, monitoring and human-in-the-loop fallback. A pure-play UK consultancy would have white-labelled the build to a third party, lengthening every feedback loop.
Xamun was the only firm in the consideration set with both: a London partner team running diagnosis and governance, and the Manila Software Factory running build. One commercial contract. One accountable team. The COO signed scope inside 11 days of first meeting.
Three phases. Twelve weeks.
XI Diagnostic
London team ran XI against the client's claims data warehouse and ops dashboards. Identified the 38% of claims that were completely deterministic (matched a policy clause without judgement), the 47% that needed pattern-match review, and the 15% genuinely complex. Built the business case: target = automate the 38% completely, AI-triage the 47%, escalate the 15%.
Software Factory Build
Four 21-day cycles in the Manila Software Factory. Cycle 1: deterministic claim classifier + audit trail. Cycle 2: LLM-driven triage agent for ambiguous claims with confidence scoring. Cycle 3: human-in-loop dashboard for the 15%. Cycle 4: integration with the client's existing case management system + monitoring.
Embed + PlayStudio
Phased rollout — 10% of inbound claims to the AI pipeline, then 40%, then 100%. PlayStudio governance dashboard live by Week 11 so the COO could see automation rate, escalation rate and override rate in real time. Two Xamun engineers embedded for handover.
What actually moved.
= 14 FTE reassigned to higher-value disputed-claim handling. Headcount preserved; productivity per FTE up 3.2×.
Payback period on the engagement: 4.5 months. ROI in year 1: 380%.
From 0 to 8,400 by week 12. Sustained at 8,200+ in the steady state. Override rate <2%.
Median claim turnaround for the automated 38% dropped from 14 minutes of human review to 38 seconds end-to-end. Member NPS up 9 points in Q1 post-launch.
All metrics independently verified by the client's internal audit team in Q1 of the following financial year.
£78k total. Outcome-priced.
The engagement was structured as one Discovery ($2,500), four 21-day Software Factory cycles ($18k–$22k each), and a two-week embed phase. Total contracted value: ~£78k. Payback period: 4.5 months from go-live.
For comparison, the client's pre-engagement pipeline included a £340k proposal from a tier-two UK consultancy for "AI strategy roadmap and PoC". That proposal would have produced a deck and a prototype. Xamun produced a £420k/year saving.
Have a claims, ops or back-office problem that looks like this?
Talk to the partner who led this engagement. Half a day Discovery in London, $2,500, walk out with an Opportunity Map for your own claims/ops workflow.