AI for underwriting, claims and service, deployed into the core systems you already run.
We have led strategy, transformation and data science inside a non-life insurer. We know claims, bancassurance, reinsurance operations, IFRS 17, telesales compliance and fraud analytics, and the difference between a demo and a system the regulator will let you run.
Where the hours go in insurance
Underwriting & new business
Document-heavy operationsSubmission intake, document extraction, risk checks, referral routing, quotation support.
FNOL, document assessment, fraud signals, reserve support, settlement drafting, with humans on the high-value cases.
Service & renewals
Customer conversations at scaleVoice and chat for servicing and renewals; review of every agent conversation instead of a sample.
Quality control
Quality, control and complianceReview every issued policy and settled claim; measured accuracy reported to the control function.
Pricing & portfolio
Decision supportAnalyst-grade first drafts for pricing reviews, exposure monitoring and reinsurance packs.
Finance & reporting
Internal operationsClose support, IFRS 17 data preparation, board and regulator packs.
Things we plan for from the diagnostic, not the go-live.
- Regulatory expectations [OIC / local regulator] on outsourcing, data and automated decisions determine deployment options; we design for them from day one.
- Distribution partners (bancassurance, brokers, agents) shape the workflow as much as the insurer does.
- Claims is where the money is and where a wrong answer is most visible. We start with augmentation, not automation, and earn the right to go further.
Our first engagements are in progress. We publish results only with the client's name and permission.
Not in insurance? Most of these workflows have a twin in your industry. See the problems we take on
Bring us the workflow that costs you the most.
Tell us what it is, who does it today, and roughly how many hours it takes. We'll come back within a week with a view on whether AI changes it, and how we'd prove it.