Friendly
The intelligence layer for insurance decisions you can defend.
Friendly reads every page, surfaces what matters and hands your underwriting and claim experts the findings ready for decisioning.
Life & health underwriting
Complex medical cases
Risk triaging
Straight-through processing
Post-issue sampling
Claims
Life claims
Long-term disability
Critical illness
Group claims
Caspar David Friedrich, Riesengebirge landscape with rising fog, 1819. Bayerische Staatsgemäldesammlungen.
SOC 2 Type II
HIPAA
GDPR
ISO 27001
Data residency in EU, UK, US & ZA
Used by
Swiss Re
PPS
Prudential
Unum
and 4 more under NDA
Sharper risk selection, shorter cycle times, a lower expense ratio, and an audit trail that still holds up two years later.
Case summarySource PDFReady
Case
Case 41-2286Life14 docs · 1,412 pp
Sources · 14 documents
Application APS Lab results Rx history Imaging Consult notes +8More
Pages read
1,412in 12m 08s
For review
2points
Key facts
✓BMI 31.2pg 16
✓HbA1c 8.4%pg 412
!Tobacco use→ sources disagreepg 12
!Echocardiogram→ ordered Mar 2025, no result on filepg 331
History
2019T2 diabetes
2021Hypertension
2023Sleep apnoea
2025Echo ordered · pending
Records · tobacco use1 discrepancy
✗Says no1 source · declared
Applicationsigned 12 Mar 2026“Tobacco in the last 5 years?” Nopg 12
✓Says yes3 sources · in the file
Lab reportdrawn 18 Feb 2026Cotinine >1.00, positivepg 88
Physician note2 Jun 2025“Smokes 5 cigarettes a day”pg 204
Prior application2019 policyDeclared smoker at issuepg 58
✗One source against three → routed to the underwriter with pg 12, 58, 88 and 204 attached. Not decided for them.
Your cloud · eu-west-2Your keysSSO
FriendlyCase 41-2286
▸Personal details12
▾Vitals3
weight218 lbpg 16
bmi31.2pg 16
blood pressure148/94pg 204
▸Diagnoses9
▸Medications14
▸Lab results236
Your rules engineManual v14
▸Personal details§1.1
▸Vitals§4.1
▸Diagnoses§6.1
▸Medications§3.2
▸Lab results§2.3
Recommended rating
Smoker · Table 2
Underwriter confirms
Runs in your cloudYour manual, your rulesNo training on your data
The number your CFO will ask for.
Not model benchmarks. Three measurements from live carrier deployments, each one against the manual process your team runs today, which is the only comparison a steering committee accepts.
FIELD CONFIDENCE · 2,675 VALUES
THRESHOLD
TO A HUMAN
HIGH CONFIDENCE
95%
Field-level accuracy, scored
Everything below your threshold routes to a human automatically. The error rate becomes a number you manage, not one you discover in an audit or appeal.
PAGES READ PER HOUR
SENIOR UNDERWRITER
~75
FRIENDLY
7,000
93× FASTER ON THE SAME EVIDENCE FILE
7,000
Pages read per hour
A thousand-page medical file stops setting your turnaround. Cycle time becomes a service level you choose, not a constraint you absorb.
HOURS PER CASE
BEFORE
AFTER
SIX CONSECUTIVE WEEKS, ONE CARRIER
85%
Less handling time per case
The same desk clears materially more volume, and the saved hours come straight off the expense ratio. With 400,000 insurance professionals retiring by the end of this year, that is the only capacity plan that survives contact with the budget.
The layer underneath every decision.
Underwriting and claims on one auditable substrate. It reads what your people read, links every finding to the page it came from, and plugs into the pipeline you already run.
01 / 08 · Case dashboard
Identifies and sources key risks
Critical risk factors surfaced from complex medical and financial documents, each one traceable.
02 / 08 · Non-disclosures
Identifies misrepresentation
Conflicting information and synthetic data flagged automatically, on every case rather than a sample.
03 / 08 · StoryMode
Chronological case view
The full story of a claimant's care assembled from every document, in order.
04 / 08 · FocusGrid
Next best actions
The issues to resolve before a decision, each with a recommended next step, so senior judgement reaches every desk.
05 / 08 · Narrative & chat
Ask across the whole file
Questions answered across every document at once, with the source attached to the answer.
06 / 08 · Handwriting recognition
Auditable extraction
Every insight links back to its originating page. The audit trail is the product, not a report written afterwards.
07 / 08 · Exclusions check
Flag claim exclusions
Claimed conditions checked against the policy's exclusions, with the evidence behind each call.
08 / 08 · Triage
Triage at a glance
Medical and financial risk flagged in the queue, before anyone opens the file.
Friendly case dashboard with the insured's details, a HealthGrid of impairment categories and pending investigations A tobacco non-disclosure: the application says no, while a cotinine result in the lab report says otherwise StoryMode: a dated narrative of diagnoses, treatments and consults FocusGrid: issues on a claim that need the assessor's attention before a decision, each with a recommended next step A claim narrative with its timeline, next to a chat answering questions about the file Fields read from a handwritten form, one flagged at 70% confidence, beside the source page An exclusions check marking each policy exclusion as applicable or not, with key evidence from the claim form The case queue with risk flags and a high-risk summary on hover
01 Identifies and sources key risks Critical risk factors surfaced from complex medical and financial documents, each one traceable. Friendly case dashboard with the insured's details, a HealthGrid of impairment categories and pending investigations
02 Identifies misrepresentation Conflicting information and synthetic data flagged automatically, on every case rather than a sample. A tobacco non-disclosure: the application says no, while a cotinine result in the lab report says otherwise
03 Chronological case view The full story of a claimant's care assembled from every document, in order. StoryMode: a dated narrative of diagnoses, treatments and consults
04 Next best actions The issues to resolve before a decision, each with a recommended next step, so senior judgement reaches every desk. FocusGrid: issues on a claim that need the assessor's attention before a decision, each with a recommended next step
05 Ask across the whole file Questions answered across every document at once, with the source attached to the answer. A claim narrative with its timeline, next to a chat answering questions about the file
06 Auditable extraction Every insight links back to its originating page. The audit trail is the product, not a report written afterwards. Fields read from a handwritten form, one flagged at 70% confidence, beside the source page
07 Flag claim exclusions Claimed conditions checked against the policy's exclusions, with the evidence behind each call. An exclusions check marking each policy exclusion as applicable or not, with key evidence from the claim form
08 Triage at a glance Medical and financial risk flagged in the queue, before anyone opens the file. The case queue with risk flags and a high-risk summary on hover
Built for global insurance. Multilingual and secure.
Friendly is the only AI solution that can natively process multilingual cases. Certified and secure, you can safely deploy it everywhere.
CASE 52-0917 · 347 PAGES
5 LANGUAGES · ONE FILE
Ärztlicher Befundbericht
DE
212pp
Informe de laboratorio
ES
34pp
Relatório de alta
PT
58pp
Questionnaire médical
FR
12pp
Application & declarations
EN
31pp
ONE RECORD
„HbA1c-Wert: 8,1 %“ DE p.44 → HbA1c 8.1% YOUR FIELDS, YOUR UNITS
Certified
SOC 2, ISO 27001, HIPAA, GDPR
Data residency
US, UK, EU, APAC — more coming soon
Built to survive the risk committee.
Three questions decide whether automated output can carry a regulated decision. A general-purpose model with a good prompt answers none of them, which is exactly where most pilots stall.
HBA1C
8.4%
P. 412
Provenance, per value
Each extracted field carries a link to its source page. An auditor can retrace any decision without reopening the file.
ROUTE TO HUMAN
STRAIGHT THROUGH
196 FIELDS
THRESHOLD 0.82 · YOURS TO SET
Confidence you can act on
Uncertain output is flagged before it reaches an underwriter, which is what makes risk-based review possible at all.
SUBMISSION
1,412 PP.
GENERAL MODEL WINDOW
TRUNCATED
Files at real scale
Multi-thousand-page submissions are processed whole. Nothing is silently truncated to fit a context window.
“Friendly's platform delivered 85% efficiency gains in our underwriting operations while maintaining the accuracy our actuaries require.”
Insurance executive, MunichRE
Run your own numbers
Two inputs, and the case makes itself.
Cases per year
Avg pages per case
Annual labour recovered
{{ recovered }}
Cost per case
{{ costPair }}
Experts redeployed
{{ hires }}
85% efficiency improvement, 7,000 pages/hour, $75/hour loaded labour rate.
Bring three cases you argued about. We'll run them and walk you through the reasoning.
Your documents, your manual, scored against your own ground truth, in week one, not after procurement.
Book a demo
Friendly
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