1 ready-to-build workflow

AI agent workflows for Insurance

Automate FNOL intake and document collection — adjusters decide, not chase.

Each recipe below is expressed only in VegaDūta's real workflow building blocks — Trigger, Agent, Knowledge, Condition, Approval, Output, Tool (MCP), Voice, Device, Loop, Code and Parallel — so it maps 1:1 to something you can assemble in the Workflow designer. Consequential actions always pass a human Approval step.

1. First-notice-of-loss (FNOL) intake & document collection

A claimant calls or messages to report a loss; an agent manually captures details, then emails back and forth for photos and documents over days.

Trigger: WhatsApp / Call / Webhook

How the workflow runs

  1. Trigger

    Loss reported. A WhatsApp, Call or Webhook trigger opens the claim in the claimant's own words/language.

  2. Agent

    Capture structured FNOL. An Agent node collects policy no., date, description and severity into a clean record.

  3. Output

    Request exactly what's needed. An Output node asks for the specific photos/documents and accepts them right in the chat.

  4. Condition

    Completeness check. A Condition node won't advance the claim until the mandatory documents are in.

  5. Approval

    Adjuster reviews. An Approval node hands a complete, structured claim to the adjuster for the decision.

  6. Tool (MCP)

    Create the claim record. A Tool (MCP) node writes it into the core claims system.

Channels & connectors

  • WhatsApp
  • Voice / Call
  • Sarvam (multilingual)
  • Approval / HITL
  • Tool (MCP → claims)

Outcome

Claims arrive complete and structured; adjusters spend time deciding, not collecting.

Why it helps

Removes the days of manual document chasing from the front of every claim, while the assessment stays with a licensed human.

Build spec

1 agent1 FNOL intake agent; completeness is a Condition node and the assessment is an Approval node.

System prompt (paste-ready)

You are a first-notice-of-loss (FNOL) intake agent. In the claimant's own language, capture a structured claim {policy_no, loss_date, loss_type, description, severity, location, injuries: boolean}. Then request exactly the documents/photos required for this loss type and accept them in the chat. Capture what the claimant reports — never assess fault, coverage, or claim value; that is the adjuster's decision. Verify the policy number format but don't approve or deny anything.

MCP connectors

  • Core claims system — via a Tool (MCP) node
  • WhatsApp / Voice — intake channel
  • Sarvam — multilingual support

Built-in tools

  • knowledge_search (required documents by loss type)
  • file parsing (uploaded photos/documents)

Guardrails

  • The Condition node blocks the claim from advancing until all mandatory documents are in
  • The agent captures only — coverage, fault and value are decided by a licensed adjuster at the Approval node
  • The policy number is validated for format; no coverage determination is made or implied

Output & delivery

The agent captures a structured FNOL and collects the required documents in-chat; the Condition node enforces completeness, the adjuster reviews at the Approval node, and a Tool node writes the claim into the core claims system.

Related industries

Build one of these in minutes

The sandbox gives you a live agent workspace — no account, no card. Or head back to the full catalogue and compare patterns across every industry.

See how VegaDūta compares to n8n, Dify and BotpressWhy VegaDūta's architectureEstimate your WhatsApp API costs