AI for insurance: agents for claims intake, submission triage and policy servicing
AI for insurance is used in practice to take first notice of loss, extract data from ACORD forms and loss runs, triage underwriting submissions, answer coverage and billing questions from policy documents, and summarize claim files. Goldenroz builds AI agents for carriers, MGAs and agencies that work inside your policy, claims and agency management systems, with a licensed person approving every coverage or payment decision. Projects start at $25,000, fixed price per phase after a free one-hour scoping call.
From $25,000, fixed price per phase
Updated 2026-11-14. By Shai Goldenberg
What usually goes wrong
- Submissions arrive as email attachmentsA commercial submission is an email with ACORD applications, loss runs, schedules and a broker's note. Underwriting assistants re-key it before an underwriter can even decline it.
- Catastrophe surgesA hailstorm or hurricane sends weeks of claim volume in a few days. First notice of loss queues grow and state deadlines for acknowledgment keep running.
- Claim files are hundreds of pagesAdjusters read medical records, estimates, photos and correspondence to find the few facts that matter. Handoffs between adjusters repeat the reading.
- Renewal and enrollment peaksJanuary 1 renewals and annual enrollment periods concentrate servicing calls about ID cards, certificates and billing into a few weeks.
- Core systems resist changeGuidewire, Duck Creek, Applied Epic and Vertafore platforms hold the record. Data leaves them through specific APIs and integration layers, not open database access.
What is included
- Line-of-business scopeA written scope for one process and one line, such as submission intake for small commercial property, with the decisions reserved for underwriters or adjusters listed.
- Form and document extractionReading of ACORD forms, loss runs, schedules of values and estimates into structured fields, each linked to its place in the source document.
- System connectionsReads and drafted writes to your policy administration, claims and agency management systems through their supported interfaces. See AI integration.
- Appetite and rule checks in codeUnderwriting appetite, authority limits and required fields are written as plain rules that run the same way on every submission.
- Licensed-person approvalCoverage positions, reserve changes, payments and declinations wait for an adjuster or underwriter. The agent prepares and never decides.
- Decision recordA log of what the agent read, what it produced and who approved it, kept in a form that supports a market conduct exam.
Where insurance AI agents fit in claims and underwriting
Insurance runs on documents written by other people: brokers, claimants, contractors, doctors. Insurance AI agents are useful at the point where those documents enter the company. They read the submission, fill the intake screen, check it against appetite and tell the underwriter what is missing. In claims they take the loss report, set up the file, request the standard documents and write the summary the adjuster would otherwise assemble.
Pricing, reserving and fraud scoring are actuarial and statistical work. They use models validated in a different way, and we do not build them. An agent may hand a clean, structured file to those models.
A language model can misread a policy form, and a wrong coverage statement to a claimant has consequences. Customer-facing answers come from the policy documents on file with the source shown, and anything outside them goes to a person.
Regulation of AI for insurers and what the build documents
Insurance is regulated state by state. Unfair claims settlement practices laws set deadlines for acknowledging and deciding claims, and unfair trade practices laws prohibit unfair discrimination in underwriting. The NAIC has published a model bulletin on insurers' use of AI systems that expects a written governance program, risk controls and oversight of third-party vendors, and a number of states have adopted it.
That bulletin puts responsibility on the insurer, including for systems built by others. We build so you can meet it: documented purpose and limits, test results, human review points, and logs. Customer data falls under Gramm-Leach-Bliley privacy and security requirements, and health information in claim files may bring HIPAA into scope.
Goldenroz writes and documents the technical controls. We are not actuaries, auditors or regulatory counsel. The delivery method is described on the AI agent development page.
How the work runs
Scoping call
Before the projectWe map the process, the systems involved and the cost of a wrong action. You get a fixed price for the first phase.
Prototype
About 1 to 2 weeksA working agent on one narrow task, run against real examples in a test environment. You see where it fails before you pay for production.
Production build
Roughly 4 to 8 weeksIntegrations, permissions, fallbacks, approval steps, logging and the evaluation set. The range depends on how many systems the agent connects to.
Launch with limits
At launchThe agent starts on a share of the volume, or with every action approved by a person. Limits widen as the logs show it holds.
What it costs
Every phase has a fixed price agreed before work starts. Model usage is paid to the model provider from your own account.
- Focused agent or agentic workflowFixed price per phaseFrom $25,000
- Larger AI platformsSeveral agents, many integrations or a full product$100,000 and up
- Scoping callEnds with a written scope and a price$0
- Model usageBilled to your own accountPaid to the provider
Who it is not for
- Automated claim denials, declinations or pricing by a language model.
- Fraud scoring, reserving or rating models. Those are actuarial and data science projects.
- Agencies with budgets under $25,000.
Who does the work
Shai leads the technology side: business software, SaaS and mobile apps, payments and escrow systems, Postgres security, AI agents built with Claude Code and MCP, and IT and infrastructure projects with senior infrastructure engineers.
Questions buyers ask
What is AI for insurance used for today?
Intake and reading. The common deployments are first notice of loss, submission extraction and triage, claim file summaries, and servicing questions about billing, certificates and coverage. Decisions on coverage, price and payment stay with licensed staff.
How much do AI agents for insurance cost?
Projects start at $25,000 at a fixed price per phase after a free one-hour scoping call. Programs that span claims and underwriting with several core system integrations run $100,000 and up. Figures are published on the pricing page.
Can an agent deny a claim or decline a submission?
No. We do not build agents that make adverse decisions. The agent can flag that a submission is outside appetite or that a document is missing, and a licensed person makes and signs the decision.
Do your AI solutions for insurance work with Guidewire, Duck Creek or Applied Epic?
They connect through the interfaces those platforms support for your license and version. We confirm the route with your IT team or the vendor during scoping, since it determines both timeline and price.
Is this suitable for an independent agency?
For larger agencies and MGAs with enough volume, yes: certificate requests, renewal data gathering and servicing email are common starting points. Small agencies are better served by features in their agency management system.
Guides on this topic
- MCP does not replace REST APIs. It wraps them so AI applications can discover and call them. Compare MCP vs API and see when an MCP server is worth building.
- MCP is an open standard that connects AI applications to your tools and data. Learn how hosts, clients and MCP servers work, and what to secure first.
- An AI agent takes actions across your systems. A chatbot answers questions. Compare the two, see examples and use cases, and use the guide to pick one.
More for insurance
AI by industry
- AI agent development
- AI for accounting
- AI app and software development
- AI code review and AI coding training
- AI strategy and readiness
- Claude consulting and implementation
- AI for construction
- AI for education
- AI for enterprise
- AI for finance and banking
- Generative AI and LLM development
- AI for healthcare
- AI for law firms
- AI for logistics
- AI for manufacturing
- MCP and Claude integration
- AI for real estate
- AI for retail and ecommerce
- AI for small and mid-size business
- AI
Talk to the person who will do the work
One hour, no sales team. You talk to Shai Goldenberg, the CTO, about your systems or your product, and you leave with a clear next step and a price range.