Skip to content
DT Nova
Business process automation

Insurance claims automation that leaves the decisions with your handlers

Insurance claims automation takes the collecting out of claims handling: the first notice of loss is recorded in minutes, documents and photos are read and checked, cover is verified against the policy, the claim is triaged to the right track and missing items are requested automatically. Handlers start from a complete file and make the decisions. We build claims processing automation for insurers, managing general agents and claims administrators around the systems they already run, hosted in the EU, and the company owns the code.

  • You see the result before it goes live
  • Fixed price for the agreed scope, in writing before work starts
  • Source code, domain and accounts in your company's name

Request a quote

A claim, before and after

A household water damage claim often goes like this today: the customer emails or fills in a web form, a handler retypes the details into the claims system, looks up the policy in another system, asks for photos, invoices and a repair estimate by email, waits, chases, reads each attachment, checks the deductible and limits, and finally decides. Most of that time is spent assembling the file, not assessing it, and the customer keeps asking where the claim stands.

StepManual todayAutomated
First notice of lossFree-form email or form, retypedGuided report by web or app, with the policy found from the customer's details
DocumentsRequested and chased by emailRequired items listed per claim type, requested and reminded automatically
ReadingEvery attachment read by a handlerInvoices, estimates, reports and receipts extracted with a confidence score
Cover checkPolicy looked up by handCover, deductible, limits and exclusions checked against the policy record
TriageQueue orderTrack chosen by your rules: fast, standard, specialist or investigation
Decision and paymentHandler decides and paysHandler decides with a proposal and reasons; payment prepared for release

Claims are one of the processes in our business process automation work. For new policy, broker and claims systems rather than automation around existing ones, see insurance software development.

What we build

Where the human review sits

The line we draw is simple: the software gathers, checks and proposes; a person makes any decision that refuses or reduces a payment. That follows from the GDPR, under which a person has the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning them or similarly significantly affects them; where an exception applies, the controller must at least provide the right to obtain human intervention, to express their point of view and to contest the decision (Regulation (EU) 2016/679, Article 22). Claims often involve health data, which Article 9 treats as a special category, so access to it is limited to the handlers who need it and logged.

Automatic acceptance can still be right for simple claims, if your rules and your legal review allow it. Fraud indicators flag; they do not reject. And every proposal shows its reasons, so a handler can disagree with it in one click and the disagreement is recorded for tuning.

The rules claims handling has to meet

(Legal texts read via the EU Publications Office on 12 November 2026.) In the United States, claims handling deadlines are set state by state, often based on the NAIC model act on unfair claims settlement practices; we configure the deadlines your legal team gives us rather than assume them. For the AI governance side, see our AI governance guide.

Accuracy, honestly

A repair estimate from a large contractor and a photographed handwritten receipt are very different documents, so no single accuracy figure means much. We measure on your claims: a month of real files runs through the test version, and you see per document type and field how often the value was right, how many claims needed a handler for reading, and which checks failed and why. Values below your threshold always go to a person. Corrections are logged and used to improve the extraction, and the measurement is repeated after each change.

Claims platforms or custom automation?

Insurers run claims on platforms such as Guidewire ClaimCenter, Duck Creek, Sapiens and Insurity, often with specialised tools for document reading, damage estimation or fraud scoring. Replacing those is rarely the point. Custom automation fills the gaps around them: the intake, the document flow, the checks and the handler workspace that the core platform does not cover the way you work.

SituationPlatform module or point toolCustom automation around your systems
Standard flow already in the core platformUse what you haveNot needed
Several legacy claims systemsSeparate tools per systemOne intake and workspace across them
Rules specific to your lines and partnersConfigurable up to a pointYour rules, versioned and readable
Explainability for the supervisorDepends on the vendorEvery check and proposal with its reason
Data location and ownershipThe vendor's platformYour account, EU data region

How we build it

  1. Walk the claims process. One line of business from first notice to payment, with handlers and team leads, including the unwritten rules.
  2. Scope and fixed price in writing. Claim types, rules, integrations and what the first phase leaves out, with a fixed price for that agreed scope.
  3. A test version on real claims. A month of closed claims runs through intake, extraction, checks and triage, and the proposals are compared with what handlers decided.
  4. Pilot and acceptance. One team uses it on new claims with every proposal reviewed, and the phase is accepted against the written scope.
  5. Roll out and measure. Time to complete file, time to decision, claims reopened and customer contacts asking for status, compared with the numbers recorded at the start.
  6. Maintenance. Updates, monitoring, rule changes and new claim types on a monthly plan you can cancel with one month's notice.

As a standard term, the source code sits in a repository in your company's own account, and the database and hosting are in your name in an EU data region. The document side is covered in more depth by document workflow automation, and customer onboarding checks by KYC automation. See prices.

Questions we get

What is insurance claims automation?

Software that takes over the routine work of a claim: recording the first notice of loss, collecting and reading documents and photos, checking the policy and cover, triaging the claim to the right handling track, requesting missing items, preparing the settlement and keeping the customer informed. Claims handlers spend their time on assessment and decisions instead of collecting information.

Can AI decide claims on its own?

In our builds, every decision that refuses or reduces a payment is made by a person. The GDPR gives people the right not to be subject to a decision based solely on automated processing that significantly affects them, with narrow exceptions and a right to human intervention. In our builds the software gathers, checks and proposes; a handler decides and the reasons are recorded.

Which claims are suited to straight-through processing?

Simple, low-value claims with clear cover and complete documents, such as some travel, device or small property claims, where your own rules allow payment after automated checks. Which claim types qualify, up to what amount and under which conditions is a business decision we write down with you, and the rules are versioned.

How is fraud handled?

Fraud indicators flag a claim for review; they never reject it. Indicators can include duplicate documents, edited images, repeated bank accounts across unrelated claims or timing patterns your investigators define. The flag sends the claim to a person with the reasons listed, so the decision stays explainable to the customer and the supervisor.

Does it work with our policy administration and claims systems?

Yes. The automation reads policies, cover and claim history from your policy administration and claims systems through their interfaces and writes the claim file, tasks and payments back. Where a core system has no interface, a robot can bridge the gap until one exists.

How accurate is the document reading?

It depends on your documents, so we measure it on them. A month of real claims runs through the test version, and you see per field and per document type how often the extracted value was right and how many claims needed a handler. The threshold for automatic acceptance of a value is set by you.

Who owns the system?

Your company. The source code sits in a repository in your own account, and the database and hosting are in your name in an EU data region. You can keep us for maintenance, run it yourself or hand it to another team.

Get a written quote

Tell us in a few lines what you need. We reply within one working day with a written quote: scope, timeline and a fixed price or an hour estimate. Asking is free and commits you to nothing.

See every price · Contact details

Request a quote