unthinkable
Insurtech · AI Product Engineering

Help people get the cover they need, and get paid when they claim, without chasing anyone.

For companies building insurance products

We help you build the product that makes it possible: AI that explains quotes, collects documents and keeps every claim moving, with the rules that must be exact running as code.

The claims work a product can take on

35%of UK car insurance complaints the ombudsman decided went the customer's way: claim files a product can keep complete and explainedFinancial Ombudsman Service, 2025/26
38%of UK buildings insurance complaints the ombudsman decided went the customer's way: decision letters a product can check against the policy wordingFinancial Ombudsman Service, 2025/26
10%of property and casualty claims are fraudulent, by Deloitte's estimate: exact checks a product can run so honest claims don't waitDeloitte Insights, 2025
What we build

What we build for insurtech companies.

The insurance products we build, each helping people buy cover, change it and get paid on a claim. For each: what changes for your customers, what AI does (and what people still decide), and how we help you build it.

Insurance marketplaces & operations

People find and buy the cover they need, and get paid when they claim, without chasing.

BUILD AI IN

Quotes explained in plain words, documents collected, renewals and changes handled, claim status shared at every step. Underwriters and adjusters make every decision.

BUILD WITH AI

Connected to insurer rating, policy and claims systems; product rules run as code.

Shipped forVimo, Seopa, Insurance Masters, ICEA Lion Life, Iffco-Tokio

Running an insurer? Insurers →

Your customers' journey

The same policyholder journey, as your product sees it.

Seven steps where policyholders wait on an insurer. At each: what your product does, what it has to connect to, and what an insurer's claims, compliance and risk teams will check before signing.

Today: a wait before every stepWith AI at work: straight through

1 · Getting a quote

Your productTurns a plain-words request into a rated quote: checks eligibility, asks the missing rating questions and explains cover and exclusions from the policy form.

Connects toThe rating engine, the policy forms library and the customer record.

Buyers checkThat every premium comes from the rating system, and the product never recommends cover in a way that needs a licence.

Insurance marketplaces & operations ↓

2 · Buying the cover

Your productPre-fills applications from the customer record, asks underwriting questions by product, reads uploaded IDs and documents and matches them to the application.

Connects toPolicy administration, identity (KYC) checks, vehicle or property data, and the document store.

Buyers checkThat underwriting rules run as code, cover is bound only by the policy system, and referrals reach an underwriter with the file complete.

Insurance marketplaces & operations ↓

3 · Changes & renewal

Your productExplains renewal premiums from recorded rating factors, takes endorsement requests in plain words, re-rates them and asks the policyholder to confirm.

Connects toPolicy administration, the rating engine, billing and the document archive.

Buyers checkThat no premium is discounted or changed outside the rating system, and every endorsement is logged with who confirmed it.

Insurance marketplaces & operations ↓

4 · Reporting a claim

Your productCaptures first notice of loss in plain words, checks policy status and cover, reads photos and documents, and opens the claim with an adjuster assigned by rule.

Connects toClaims platforms such as Guidewire or Duck Creek, the policy record and the document store.

Buyers checkThat injury or danger always reaches a person first, and the product never says whether a claim will be paid.

Insurance marketplaces & operations ↓

5 · The assessment

Your productExplains each claim's next step and owner in plain words, books entitlements such as rentals within the policy, and runs exact checks for duplicates, dates and linked parties.

Connects toThe claims platform, the policy record, repair and rental networks, and fraud rules or models.

Buyers checkThat a fraud signal is never an accusation, and the product never tells a policyholder a claim is covered or denied; the adjuster does.

Insurance marketplaces & operations ↓

6 · Getting paid

Your productCalculates the settlement from the approved amount, limits and deductible, explains it from the policy and tracks payment to the policyholder or repairer.

Connects toThe claims platform, the payment rails and the policy record.

Buyers checkThat no money moves until an adjuster approves the amount and a named person releases it.

Insurance marketplaces & operations ↓

7 · Questions & complaints

Your productExplains decisions from the issued letter and policy wording, collects new evidence and opens the complaint with its regulatory deadline.

Connects toClaim files, issued policies, the complaints queue and regulator correspondence.

Buyers checkThat the product never reverses a decision or admits fault; a reviewer does, and every step is logged.

Insurance marketplaces & operations ↓

What the product does at each moment (text version)

Getting a quote

  • Understand the request (Plain words, any channel) → Eligibility and rating (Underwriting rules, rating system) → Explain the quote (From the policy form only) → Licensed agent advises (Which cover to buy). Guardrail: The agent never sets a premium or tells a customer which cover to buy; a licensed agent advises.

Buying the cover

  • Understand the request (Plain words, any channel) → Pre-fill and ask (Customer record, underwriting questions) → Read the documents (IDs, cards, photos) → Check against the application (Exact field matches) → Underwriter decides (Anything outside the rules). Guardrail: The agent never binds cover or promises a premium; an underwriter decides and the policy system issues the policy.

Changes & renewal

  • Policyholder asks (App, text or web chat) → Read the rating factors (Policy system) → Explain the change (Plain words) → Re-rate the endorsement (Rating system) → Licensed agent or underwriter (Anything outside the rules). Guardrail: The agent never discounts a premium or advises which cover to choose; changes outside the rules go to a person.

Reporting a claim

  • Policyholder reports (App, phone or text) → Check the policy (In force, cover on the vehicle) → Collect the details (Photos, other driver, what happened) → Open the claim (Adjuster assigned by rule) → Adjuster takes it (Every claim). Guardrail: Any mention of injury or danger goes straight to a person, and the agent never says whether the claim will be paid.

The assessment

  • Policyholder asks (Any channel) → Read claim status (Claims system and checks) → Explain in plain words (Next step and who owns it) → Act within policy (Rental, appointments) → Adjuster decides (Coverage and payment). Guardrail: The agent never tells a policyholder a claim is covered or denied; the adjuster does.

Getting paid

  • Adjuster approves (The amount) → Calculate the payment (Limits and deductible) → Explain it (From the policy) → Track the payment (To the policyholder or repairer) → Payments releases money (Final check). Guardrail: No payment goes out until an adjuster approves the amount and a person on the payments team releases it.

Questions & complaints

  • Policyholder disputes (Any channel) → Read the letter and clause (Issued policy) → Explain the decision (Plain words, from the letter only) → Open the complaint (With evidence and deadline) → Claims manager decides (Every complaint). Guardrail: The agent never reverses a decision or admits fault; a claims manager reviews every complaint.

Try it with your own case →

Why products stall

Why insurance products stall, and how we help you avoid it.

BUILD AI IN
“It will tell a customer their claim is covered.”
Coverage and payment stay with adjusters. The language model explains the claim's stage, the policy and the letter; it never decides or hints at an outcome.
“A regulator will ask how it decided.”
Every input, model version, prompt, output and human approval is logged, so an examiner can replay how a quote, endorsement or claim step was reached.
“It will price or treat some customers unfairly.”
Premiums come from the filed rating rules, not the model. We test AI outputs across groups, document model behaviour for the insurer's governance process, and monitor drift after launch.
BUILD WITH AI
“It won't fit the insurer's core systems.”
We connect to rating engines, policy administration and claims platforms such as Guidewire or Duck Creek through their APIs, sandbox first, with least-privilege access.
“Policyholders' data will train someone's model.”
We never train models on your data, and we configure model providers so it is not retained or used for training. Data stays in your cloud account.
“Nothing in insurance moves quickly.”
We start with one workflow as a fixed piece, such as claim status or first notice of loss, built sandbox-first with evaluation sets and approval steps agreed with the insurer's risk team, then extend once it holds.
You inherit

Building blocks already built.

Already built
Document extraction for IDs, insurance cards, estimates and claim photosin production
Decision log with replayin production
Policy-clause and rating-factor explanation layerin production
Human approval queue with limits by rolein production
PII redaction, ID masking and encryptionin production
Evaluation suite for extraction accuracy and fairnessin production
Your insurtech productstarts on these, not from scratch

The team around it

Product designerShapes how people and agents work together
AI engineerAgents, prompts, evaluations
ArchitectRecords, integrations, security
DevelopersBuild it with AI, keep it maintainable
TestersSpecs and tests that prove it works
Domain expertKnows your industry's rules and edge cases
Security reviews

Will your product pass an insurer's compliance and AI governance review?

Before a pilot, an insurer sends its vendor questionnaire and asks how your AI is governed, tested and logged. These are the answers we build in from the first release.

BUILD AI IN
A decision log an examiner can replay
Every input, model version, prompt, output and human approval is stored against the quote, endorsement or claim step it produced.
Wording from the policy, not from the model
Premiums come from the rating system, limits and clauses from the issued policy, and payments from code; the language model only explains them.
Governance evidence
AI outputs are tested before release, model behaviour is documented for the insurer's AI governance process, in the spirit of the NAIC model bulletin on insurers' use of AI, and drift is monitored after launch.
BUILD WITH AI
Data that stays put
Data stays in the buyer's or your cloud account where required, encrypted in transit and at rest; model providers are configured not to retain it or train on it.
Least-privilege integrations
Each integration runs with its own credentials and access limits, and every call that issues cover, changes a policy or moves money is logged.
Controls auditors already use
Built to the GLBA and GDPR controls the buyer's auditors work with, and tested in vendor sandboxes before any live data.
Integrations

The integrations insurers ask for.

An insurer will ask what your product connects to before it asks about features. These are the connections we build.

Rating and policy administration

The insurer's rating engine and policy administration system, through their APIs, sandbox first.

Claims platforms

Guidewire, Duck Creek or an insurer's own claims system, plus repair, rental and inspection networks.

Documents and identity

Document capture and content systems, identity (KYC) checks and ID masking in stored documents.

Payments and benefits

Payment systems for premiums and claim payments, and partner services such as doctor consults or lab bookings, each with its own credentials.

You own the result: code, prompts, evaluation sets and data sit in your accounts from the first line, so your product's IP stays yours.

Proof

Insurance products we have shipped, and what they did.

ACKO, India (insurtech)90+companies onboarded within months (as reported)

A rebuilt app for group health cover: policies, claims and endorsements tracked, doctor consults, lab bookings and fitness rewards. Its corporate health insurance app was slow and couldn't handle several policies per person.

Read the case study →
A leading Indian general insurer60TBmigrated with no data loss (as reported)

A content system upgrade and a migration from DB2 to Oracle 19c, with capture that reads and indexes ID cards, forms and vehicle images. Its document system had dropped support for DB2, and 60 TB of live policy documents and vehicle photos had to move without stopping the business.

Read the case study →
Pramerica Life Insurance, India98%accuracy across 500,000+ documents (as reported)

Bots and an image utility that find the ID number on each scanned document and mask it in place, on the insurer's own machines. A new government rule required national ID numbers to be stored masked, across about 500,000 hand-scanned documents.

Read the case study →

Results as stated in each linked case study; figures marked "as reported" to be checked against the source before launch.

Which step would your product take the chasing out of first? Name the step your policyholders chase most: a quote, a renewal, a claim or a payment. After a short discovery you get a plan with dates for the first release, measured against those calls.

Tap one: Ask Unthinkable starts your plan from there. Or say it in your own words ↓

How to choose

How to choose who builds an insurance platform.

Four things an insurtech founder or CTO should check, and how we meet them.

Has shipped insurance products

A group health insurance app for ACKO, a 60 TB policy-document migration for an Indian general insurer, and ID masking across 500,000+ documents for Pramerica Life Insurance (as reported).

Often elsewhereStrong engineers who are new to rating, endorsements and claims rules.

Knows the insurer's side

We also help insurers run day to day, so we know the claim deadlines, decision letters and examiner questions your product has to fit.

Often elsewhereProduct-only shops know the app; few have sat with an adjuster.

Agents inside the product

In claims, for example: first notice of loss from the scene, the claim's next step and owner in plain words, and the payment explained from the policy.

Often elsewhereA chat widget bolted onto the app.

Feature twenty as fast as feature one

We set the architecture before the prototype and keep AI writing inside it; eligibility, rating calls, limits and payment rules stay in code, so a product change is a tested release.

Often elsewhereA fast demo that stalls once core-system integrations and the insurer's governance review arrive.

The full comparison: us, other software companies, in-house teams and consultants →

Ways to work

Start the way that suits you.

Time & materials

A senior AI team billed for time, within an agreed monthly budget. Even on time and materials the work is tied to your business objective: every plan starts from it, and every report shows the number next to the hours.

One fixed outcome

One agreed outcome at a fixed price, with acceptance criteria, including the measure, written up front.

Dedicated team

A complete senior AI team, monthly, working as part of yours and reviewed each quarter on what shipped and what moved.

Linked to the result

Where the gain can be measured from your own systems: a reduced implementation fee plus a percentage of the expected gain, for a set period.

How we work and price →

Straight answers

Questions insurtech teams ask.

How do I choose a development partner for an insurtech product?

Judge them on what your insurer and distributor buyers will judge you on. Ask how their product would explain a renewal increase or a declined claim, and whether the wording comes from the issued policy and recorded rating factors or from a model. Ask to see a decision log an examiner could replay: inputs, model version, prompt, output and the person who approved. Ask how they connect to rating engines, policy administration and claims platforms such as Guidewire or Duck Creek, and whether they build against sandboxes first. Then check that code, prompts and data sit in your own cloud account. Unthinkable has shipped insurance products such as ACKO's group health app, and also helps insurers run, so it knows the questions your buyers will ask.

Can AI speed up insurance claims without increasing fraud?

It can, when fraud checks stay in code and with people. The agent takes the first notice of loss, checks the policy and cover, and gathers photos and documents, which removes most of the waiting. Rules and your fraud models score each claim, and anything unusual goes to an adjuster with a summary of why. Straightforward claims move faster; suspicious ones get more attention, not less.

Can an AI agent quote insurance without giving advice?

Yes, if the line is drawn in the product. The agent asks the rating questions, returns the premium from the insurer's rating engine and explains what the policy form covers and leaves out. Recommending which cover to buy stays with a licensed agent, and the product hands the conversation over when a customer asks for advice. Which questions need a licence depends on the states and products you sell in.

How do you build a first notice of loss agent?

We start from the claim types your insurers see most, such as auto or property, and the questions their adjusters ask first. The agent checks the policy was in force and what cover it carries in code, collects photos and details in plain words, and opens the claim with an adjuster assigned by rule. Any mention of injury goes straight to a person. We measure it against claims your insurer has already handled before going live.

How much does it cost to add AI to an insurance platform?

It depends mostly on how many systems we connect and how strict your audit requirements are, not on the AI model. Most clients start with one workflow as one fixed outcome, covering integration, evaluation and a production release, and get a written estimate after a short discovery. Unthinkable measures running cost per claim or quote during that first release, so you know the unit economics before scaling.

Is our policyholders' data used to train AI models?

No. We never train models on your data, and we configure model providers so your data is not retained or used for training. Data stays in your cloud account where required, is encrypted in transit and at rest, and access is logged. You own the code, IP and data. We work to the GLBA and GDPR controls your auditors already use.

Can you integrate with Guidewire, Duck Creek and our rating engine?

Yes. Unthinkable connects through the APIs your rating engine, policy administration and claims platforms support, such as Guidewire or Duck Creek, and builds against their sandbox environments first. Each integration runs with its own credentials and access limits, and every call that issues cover, changes a policy or moves money is logged. Exact scope depends on your vendors and agreements.

Ask Unthinkable

Which step will your policyholders stop chasing first?

Tell us about your insurance product and the moment you want to change. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with a senior AI engineer.

Send us a note instead

Required

Thanks. A senior AI engineer will reply to you.
u
Ask UnthinkableClarity first, then the right team
Your conversation

An AI agent. This prototype uses prepared answers and sends nothing.