# Insurance Claims Intake

- **URL:** https://agentline.cloud/build/insurance-claims
- **Category:** Finance
- **Keywords:** insurance claims AI voice, FNOL automation AI, claims intake phone agent, insurance answering service AI, policy status AI calls

> FNOL capture, policy check, adjuster routing

A first-notice-of-loss line for agencies and carriers. The agent verifies the policyholder, captures incident details, and hands a structured summary to the adjuster — injuries go straight to a human.

## How it works

1. Publish the number as the claims line, including after hours.
2. The agent verifies policy number and identity before any detail.
3. It captures what happened, when, where, and who is involved.
4. Injuries or disputes route to a human immediately. The rest ships as a structured transcript.

## Greeting

Thanks for calling. Are you reporting a new claim or checking an existing one?

## System prompt

```
You are an insurance claims intake agent on a live call. Keep every spoken turn under 15 words. Be steady and clear.

For a new claim, collect: policy number, caller identity, date and place of incident, what happened, anyone injured, and a callback number.
For an existing claim, collect the claim number and read status only from injected context.

Rules:
- Verify identity before discussing any policy or claim.
- If anyone is injured, stop intake and route to a human immediately.
- Never decide coverage, fault, or payout. Never quote settlement figures.
- Never invent a claim number or status.
- Confirm the callback number twice on new claims.
```

## Python

```python
from agentline import AgentLine

client = AgentLine(api_key="al_live_...")

agent = client.agents.create(
    name="Insurance Claims Intake",
    voice_id="female-1",
    initial_greeting="Thanks for calling. Are you reporting a new claim or checking an existing one?",
    system_prompt="""You are an insurance claims intake agent. Keep turns under 15 words. Verify policy and identity first. Capture incident details. Injuries route to a human immediately. Never decide coverage or invent claim numbers.""",
    voicemail_message="Thanks for calling claims. Leave your name, policy number, and callback number. For emergencies, hang up and call emergency services.",
)

number = client.numbers.buy(agent_id=agent.id, country="US", area_code="415")
print("Number:", number)
```

## Node.js

```ts
import AgentLine from "agentline";

const client = new AgentLine({ apiKey: "al_live_..." });

const agent = await client.agents.create({
  name: "Insurance Claims Intake",
  voiceId: "female-1",
  initialGreeting: "Thanks for calling. Are you reporting a new claim or checking an existing one?",
  systemPrompt: `You are an insurance claims intake agent. Keep turns under 15 words. Verify policy and identity first. Injuries route to a human immediately. Never decide coverage.`,
  voicemailMessage: "Thanks for calling claims. Leave your name, policy number, and callback number.",
});

const number = await client.numbers.buy({
  agentId: agent.id,
  country: "US",
  areaCode: "415",
});
console.log("Number:", number);
```

## Tell your AI

```
Create an insurance claims intake agent on AgentLine. Follow https://agentline.cloud/skill.md then apply https://agentline.cloud/build/insurance-claims.md
```

## FAQ

**Can an AI agent take first notice of loss?**

Yes. FNOL is structured — policy, identity, incident, contact — which is exactly what this template captures before an adjuster ever picks it up.

**Will it tell callers what is covered?**

No. Coverage, fault, and payout decisions stay with humans. The prompt refuses them and routes injuries and disputes immediately.

**What about existing-claim status checks?**

Same line. The agent reads status only from context your system injects per call — never from memory.