Built by people who read the case.
Friendly is the evidence layer for insurance decisions: multilingual, global, and built around the part of the job that consumes the day, which is reading.
We exist so that an underwriting or claims decision can be made on everything in the file, and defended on it years later, in front of whoever asks.
Founded
2022
Lines in production
Life, Disability, Employer Benefits, Workers’ Comp
Data residency
EU, UK, US & ZA
Deployment
Your cloud or ours
The lines we work in
We work four lines, and we know them properly.
Life, Disability, Employer Benefits and Workers’ Compensation: the lines where one case arrives as hundreds of pages of medical history, and where the decision has to hold up years later. That is the problem we built for, rather than claims in general.
Life
Applications, APSs and decades of medical history, read in full.
Disability
Files that keep growing for as long as the claim stays open.
Employer Benefits
Group schemes: many claimants, one set of scheme rules.
Workers’ Comp
Treatment records, causation and the road back to work.
Why we exist
Insurance runs on documents nobody has time to read in full.
A thousand pages arrive, an underwriter has an afternoon, and the decision gets made on the parts that were findable. For a carrier that gap shows up as mispricing on the way in, leakage on the way out, and exposure every time a regulator asks how a decision was made.
We built a system that reads all of it and shows its work. Not a chatbot with an insurance prompt, but an infrastructure layer that produces evidence a regulator, an actuary and a claimant's lawyer can all follow.
The outcome we sell is simple: more of the right business written, fewer of the wrong claims paid, and every decision still standing when someone comes back to it.
Three commitments that decide what we will and will not ship.
01
Show the source
No output without provenance. If the system cannot point to where a value came from, it does not belong in a decision, and it does not reach one.
02
Say the confidence
A number without a confidence score is a guess wearing a suit. Ours are scored, the threshold is yours to set, and everything below it goes to a person.
03
Prove it on your data
Benchmarks are marketing. The only accuracy figure that matters is the one measured on your cases, against your own ground truth, before you sign anything.
You should know whether it works before procurement does.
WEEK 1 · YOUR DOCUMENTS
WEEK 2 · YOUR ANSWER
Week one, on your documents
We start on your files, against your own ground truth. If the accuracy is not there on your cases you should not buy it, and you will know inside a fortnight rather than at the end of a procurement cycle.
Built to clear your security review, not to start one.
Software only, deployed in your cloud or ours, with data residency honoured per region. The controls are configuration rather than a support ticket, and the documentation is ready before the first call.
In
Ingest
Encrypted in transit and at rest, with regional processing where residency requires it.
During
Process
No training on your data. Isolated tenancy, and every read and export written to the log.
Out
Return
Structured values with citations, delivered to your systems by API or export.
After
Retain
Retention set by your policy. Deletion is verifiable, and reported when it happens.
SOC 2 Type II
HIPAA
GDPR
ISO 27001
Residency in EU, UK, US & ZA
No training on customer data
Four years of building one thing properly, in a market that does not forgive shortcuts.
2022
Founded
Started with a single question: what would it take to read every page of every submission.
2023
First carrier in production
A life book in Southern Africa, measured against the underwriters it was meant to help.
2024
Claims engine
The same substrate extended from underwriting into disability and supplemental benefits.
2025
Global reinsurer deployment
An 85% reduction in handling time, verified against their own ground truth.
2026
Distribution network
Building the partner network that puts Friendly in front of carriers worldwide, deployed alongside the people who already serve them.
Machine learning was our day job long before AI became the headline.
Natasha Alexeeva
Founder & CEO
A proven leader with over 15 years of multi-industry experience in machine learning and artificial intelligence, data analytics, program management, supplier and vendor management, and software architecture.
Charles Rentmeesters
Chief Technology Officer
25 years of experience in artificial intelligence and software engineering. An innovator and entrepreneur who excels in dance and has explored the South Pole.
Fionna Kossmann
Claims Product Owner
Brings global reinsurance expertise and a sharp eye for claims innovation. She specializes in scaling AI-powered solutions for life and health claims teams.
Mike Pado
Principal & Senior Advisor
Many years of experience in reinsurance and business strategy. Former CEO of PartnerRe and SCOR Global Life Re.
Kevin Smith
Underwriting Expert & Advisor
Decades of underwriting leadership at RGA, Legal & General, Liberty Life and Jefferson Pilot. Supports product-market fit for life underwriting AI and drives growth from ideation to execution.
Kevin Glasgow
Claims Expert & Advisor
Deep claims and risk experience from Munich Re, Swiss Re and GE Financial. Advises on the development of Friendly’s claims adjudication solution.
Used by
Swiss Re
PPS
Prudential
Unum
And four more under NDA: global reinsurers, national carriers and group benefits providers across Life, Disability, Employer Benefits and Workers’ Compensation.
Come and break it on your hardest file. We would rather find the limit together than have you find it in production.
The intelligence layer for decisions you can defend. Underwriting and claims, on evidence your auditors can follow.
Solutions
Resources
Company
SOC 2 Type II
HIPAA
GDPR
ISO 27001
© 2026 Friendly