Use cases by industry3 use cases
AI automation for insurance brokers, MGAs and claims teams
A commercial broker's account handler spends renewal season assembling submissions: the application form (ACORD in the US, a proposal form in the UK), five years of loss runs in five carrier formats and a schedule of values in Excel, then re-keys the same data into each carrier's portal and chases quotes. Insurance submission intake extracts and normalizes that data once and builds the submission per carrier appetite; the broker still advises and places.
The carrier side is well funded. FurtherAI raised a $25M Series A from a16z and reports one MGA going from about 32 minutes to about 1 minute per submission, its own figure; Sixfold raised $30M for underwriting; and Applied Systems announced a Submissions Manager inside Epic. Independent brokers on Acturis, Open GI or Vertafore, with their own carrier panels and portals, are where custom work still fits.
Much of the friction is portals. Carriers expose quotes, policy documents and claim status through web portals without APIs, so staff log into six of them a day. A computer-use agent for API-less portals can carry the data across, logging a screenshot at each step and stopping before anything is bound or sent, and document fraud detection cross-checks claims evidence against independent data.
What I would build for insurance
Automating portals and legacy systems that have no API, with a person approving the final click
Keys data into insurer portals and API-less desktop systems with a computer-use model, logs every screen, and waits for a person to approve each submission.
Broker management system / Insurer and carrier portals / A legacy desktop ERP / Computer-use model (Copilot Studio, Claude or OpenAI) / Outlook
Catching edited bank statements, payslips and invoices before they cost you money
Cross-checks uploaded statements, payslips and invoices against bank data, VAT records and the file's own structure, then puts the evidence in front of a reviewer.
Upload portal / Open banking provider / EU VIES / KYC provider (Onfido, IDnow or similar) / CRM or loan system
Commercial insurance submissions assembled from proposal forms, loss runs and property schedules
Reads a client's proposal forms, loss runs and property schedules, reconciles them, and drafts one submission per insurer for the account handler to check and send.
Broker management system (Acturis, Applied Epic, Vertafore) / Outlook / Excel schedules of values / Insurer portals and e-trading / BiPRO interfaces (Germany)
Renewal season, claims intake and the work in between
Loss runs in every format
Each carrier lays out claims, reserves and dates differently. Normalizing them into one table per client, with totals checked against the source, is the step account handlers dread and the one extraction does best.
Policy checking before delivery
Comparing the issued policy with the binder and the quote, clause by clause, and listing every discrepancy before the client sees it is an errors and omissions safeguard that is rarely done thoroughly by hand.
Certificates and endorsements
Certificate requests and routine endorsements follow rules that can be encoded; anything outside them goes to a handler. Renewal packs can be prepared weeks earlier from the same data.
First notice of loss
For claims administrators, capturing the first notice, collecting documents, checking coverage basics and routing to the right adjuster is intake work. The adjuster decides coverage and payment.
What regulation and E&O exposure change
Advice stays human
The system never recommends cover or tells a client what they need. The IDD demands-and-needs record in the EU is produced from the adviser's conversation, not by the model.
A record for the E&O file
When a placement is questioned years later, the broker has to show what was sent to which carrier, and when. Every submission, portal step and quote is logged with the documents as they were at the time.
Coverage and payment decisions
A model can summarize a claim file and flag what is missing. Whether a loss is covered, and what is paid, is decided by an adjuster, with fraud signals shown as evidence rather than a verdict.
German data standards
In Germany, BiPRO standards cover much of the data exchange between brokers and insurers. Where a BiPRO interface exists, use it; screen automation is for the gaps.
Software these builds usually connect to
The systems do not get replaced. The build sits across them, reads from them through their APIs, and writes results back.
Applied Epic / Vertafore AMS360 / Vertafore Sagitta / HawkSoft / EZLynx / Acturis / Open GI / SSP / Ameise / Guidewire / Microsoft Outlook / Excel
Frequently asked questions
Can AI build insurance submissions from ACORD forms and loss runs?
Yes, as a preparation step. It extracts the application data, loss history and schedule of values, normalizes them, checks totals against the source documents and fills each carrier's submission format or portal. The account handler reviews every submission before it goes out, because errors and omissions exposure sits with the broker, not the software.
Can an AI agent log into carrier portals for us?
It can, for the repetitive part. A computer-use agent opens each carrier portal, enters the submission from the normalized data, downloads quotes and policy documents and lines them up in a comparison for the account handler. Logins with MFA stay with a person, and it stops before binding or submitting anything. In Germany, look for a BiPRO interface first; it is sturdier than any screen automation.
Is AI in insurance high-risk under the EU AI Act?
Some of it. AI used for risk assessment and pricing of individuals in life and health insurance is listed as high-risk, with obligations that now apply from 2 December 2027 after the Digital Omnibus. Submission intake, document extraction, policy checking and claims triage that informs a human adjuster are outside that category, though GDPR and your own E&O duties still apply.
Should a broker buy FurtherAI or Applied's tools instead of building?
If you run Applied Epic and your panel fits its Submissions Manager, start there. FurtherAI and Sixfold are aimed mainly at carriers and MGAs. A custom build fits independent brokers with their own panels, portals and data formats, UK or German systems such as Acturis or BiPRO flows, or claims teams whose intake rules are specific to their clients.