Insurance & underwriting
In underwriting and claims, the exception is not the edge case. It is the job. Automation specified for the happy path moves work into the referral queue rather than out of it.
Where we get brought in.
Underwriter capacity
An underwriter spends the day chasing documents, re-keying submission data and writing up files. The risk judgement itself is maybe an hour of it. We separate the two so the hour gets more room.
Claims and referral queues
Every claims team has a spreadsheet the system does not know about. Some of those workarounds exist because the process is wrong, some because someone once preferred it that way. Sorting one from the other before the build is the difference between automating the operation and automating the mess.
Where judgement must stay, and why
Some decisions stay human because a regulator says so. Some stay human because getting them wrong costs more than the automation saves. Most stay human because they always have. We separate the three and write the answer into the architecture, not into a policy nobody reads.
Retail and commercial service design
Boards want a position on AI, not a list of pilots. Where it goes, where it does not, what you are prepared to be accountable for, and what you say to a broker who asks.
CASE STUDYFrom philosophy to a working prototype in four weeks
“
You clearly articulated the element of trust we need to establish, which I hadn't brought forward as one of the main strategic pillars. It's now pillar four of our entire vision and strategy around technology. That piece of insight has completely changed how we're doing our trials.
Chief Product Officer
Leadership at an independent specialist insurer had a conviction that technology should sharpen underwriters' judgement rather than replace it.
What they did not have was a way to turn that into something buildable.
Four weeks of discovery and design with our experts produced a working prototype and, more usefully, an articulation of the trust that had to be established for any of it to hold.
Automating the busywork around the risk decision rather than the decision itself. On target to return around 200 hours a week to underwriters, who spend it on the work their clients actually value.
200
hours per week expert time being reclaimed
“
When an expert underwriter tells us they’re at capacity, we ask what they achieved yesterday. Sometimes the answer is busywork - chasing a broker for a schedule, rekeying a survey report, or working out whether their last submission got a reply. None of that is putting their expertise to good use, which is what the client is paying a premium for, and usually what the colleague enjoys doing most. That’s how we start figuring out what can be automated and what needs to stay human.
Calum MacLeod, Head of Delivery and Performance
