Use case · Insurance claims intake
Agent Serve reads each claim notice, pulls out the policy and loss details, lists missing evidence and routes the claim to the right handler.
An insurance claims intake use case for Agent Serve, the AI workspace where teams build, deploy, and manage AI agents: a workflow reads each first notice of loss, extracts policy and loss details, checks the file against claims procedures in a knowledge base, logs it in a table by severity and waits for a claims handler, who decides cover.
At a glance
The problem it removes
Claim notices arrive by email with photos, invoices and forms attached. A handler opens each one, retypes the policy number and loss details, works out what is missing, and writes to the customer. Simple claims wait behind complex ones, and nobody can see the queue by severity.
How it runs
Agent Serve does the reading and the sorting. A claims handler checks the file and decides cover.
A new email in the claims inbox starts the workflow, and its photos, invoices and forms are saved to the claim’s folder in the Library.
An agent reads the notice and attachments and pulls out the policy number, date and type of loss, and the estimate if there is one.
The Knowledge block finds what the claims procedures need for this type of loss, and a Condition step checks the file is complete.
The claim is logged by severity in the claims table, and a Human step hands the file and a draft evidence request to a handler.
In depth
Agent Serve gives handlers a complete file, the policy wording and a queue they can sort.
A handler asks Agent Serve Chat whether a loss is covered, and the answer quotes the policy wording with a link to the document.
New claims land as rows in an Agent Serve table, and a workflow column classifies the severity of each one before a handler picks it up.
In the Agent Serve workflow builder each step says what it does, so a claims lead can see why a claim went to fast track and change the condition.
Agent Serve Logs keep each block’s inputs and outputs for every run, and the workspace audit log records changes to tables and workflows.
Before and after
| Work | Before | With Agent Serve |
|---|---|---|
| Reading the notice | BeforeA handler opens each email and retypes the details. | With Agent ServeAgent Serve pulls out the policy and loss details on arrival. |
| Missing evidence | BeforeFound on the first look, often days after the claim came in. | With Agent ServeListed against the procedures as soon as the claim arrives. |
| Simple claims | BeforeWait in the same queue as complex ones. | With Agent ServeGo to fast track by a condition the team can read. |
| Seeing the queue | BeforeCounted by hand from the inbox and a spreadsheet. | With Agent ServeA dashboard shows claims by severity, age and handler. |
What does the work
A workflow reads each claim, checks it and routes it by severity.
Photos, invoices and forms are saved to the claim’s Library folder.
Policy wordings and claims procedures sit in a knowledge base.
A claims table holds every claim with its severity and handler.
A dashboard shows open claims by severity, age and handler.
Examples
Three claims processes insurers run on Agent Serve, each with a handler deciding cover.
A new claim starts an Agent Serve workflow that extracts the policy and loss details, checks the procedures and routes simple claims to fast track.
Handlers drop photos, invoices and adjuster notes into one Agent Serve folder, and an agent lists what is still missing.
An Agent Serve workflow compares each claim decision with the claims procedures and flags notes the procedure does not support.
Examples show what teams can build in Agent Serve. They describe patterns, not named customer deployments.
Governance
Agent Serve keeps the cover decision with a handler and keeps claimant data to the people who need it.
Security and governanceThe Agent Serve Human block pauses each claim until a handler responds, and can alert them in Slack or by email.
Agent Serve row limits and restricted columns keep claimant details to the handlers and teams you name.
A Guardrails block detects personal data and blocks or masks it before text reaches the next step.
Every workflow run records each block’s inputs, outputs and errors, so a reviewer can trace how a result was reached.
Proposals, approvals, declines and executed actions are written to the workspace audit log, which admins can export.
Admin, write and read roles per workspace, plus permission groups that limit which models, blocks and integrations a team can use.
Sign in through your identity provider with SAML 2.0 or OIDC, and provision people with SCIM.
Run Agent Serve in your own cloud or data centre, next to the systems it reads.
Zero data retention with supported model providers, backed by your own provider keys and agreements or self-hosted models. Some provider features keep data by design, such as Gemini Deep Research, so they are outside zero data retention.
Workflows run from a deployed version, so a change is reviewed before it reaches production runs.
Further reading
A new claim arrives as an email, a form and a few photos. Here is how three workflows turn that into a structured file, a policy check and a routed queue, while every cover and settlement decision stays with a claims handler.
Insurance · 5 min readCommercial underwriters lose hours to an inbox of broker submissions before any risk is assessed. This long read argues for a Relay that reads, sorts and briefs, while underwriters keep every decision.
Commercial insurance · 5 min readOperations staff ask the same policy questions every day, and a wrong answer becomes a breach. This guide compares four ways to answer them, and shows when a cited, logged Q&A workflow is the right choice.
Banking operations · 6 min readQuestions
No. Agent Serve reads, checks and routes claims. A claims handler decides cover on every claim at a Human step.
An Agent Serve workflow can start from a new email in Gmail or Outlook, a web form through a webhook, a new row in a table, or an event from a connected app.
Agent Serve workflows can read documents with blocks such as AWS Textract or Mistral Parser, and Agent blocks can take image files as input.
Agent Serve row limits and restricted columns decide who sees each claim and which fields, in tables, dashboards and exports.
A team can start with the Agent Serve intake workflow, a claims table and a knowledge base of procedures, then add severity routing and the claims dashboard.
Last updated
Enterprise access
Agent Serve is available to teams by invitation. Tell us what you want to build and we will set up your workspace with you.