What a P&C insurance AI agent actually does
Reads a quote request, a claim or a policy-change call, pulls the right policy and carrier data, and either completes the task or routes it to a licensed agent.
A P&C insurance AI agent sits in front of an agency's phone line, website or claims intake and runs a loop: read the caller's request, pull policy, quote or claim data from the agency management system, decide whether it can complete the task without a license, and either act or route to a licensed agent or adjuster. Sonant is built specifically for P&C agencies, handling quote intake, first notice of loss and call routing, and became an Applied Certified Vendor for Applied Epic in August 2026; it also integrates with EZLynx, HawkSoft and AMS360. Liberate's voice agent, named Nicole, handles both inbound and outbound sales and service calls for insurers and raised a $50M Series B in October 2025.
Liberate's vendor-reported results illustrate why carriers are moving fast here: one insurer reported a 263% ROI with sales up 15% and operating expense down 23%, and another reportedly handled 14 times its normal claim volume during a hurricane surge without adding staff. These are vendor claims, not independently audited figures, but they explain why catastrophe-surge and after-hours coverage are the clearest wedge for AI agents in this industry.
Where agents still fail: anything that counts as binding coverage or giving advice on what to buy, which in most states only a licensed agent can do, and any claims conversation that needs human judgment about liability or fraud. TCPA's one-to-one consent rule for shared insurance leads was vacated by the 11th Circuit in January 2025, but the ordinary TCPA consent standard for AI voice calls, in force since the FCC's February 2024 ruling, still applies, and getting that wrong is one of the fastest ways an outbound lead-gen agent turns into legal exposure.
AI capabilities involved
Quote intake and initial underwriting data capture
First notice of loss and claims status
Policy-service changes (add a driver, a vehicle)
Outbound renewal and cross-sell calling
Who uses this
- Independent P&C agencies fielding high call volume for quotes, FNOL and policy changes
- Carriers wanting after-hours and catastrophe-surge claim intake without adding seasonal staff
- Agencies already on Applied Epic, EZLynx or HawkSoft deciding whether to add a certified AI layer
- Insurers running outbound renewal and cross-sell campaigns that need TCPA-safe consent handling
What the P&C insurance agent handles (and what it shouldn't)
| Task | Agent handles it? | Systems it touches | Human role |
|---|---|---|---|
| Quote intake | With approval | AMS (Applied Epic, EZLynx, HawkSoft, AMS360) | Reviews and binds the policy |
| First notice of loss (FNOL) | Fully | AMS, claims system | Reviews anything involving injury or large loss |
| Claims status updates | Fully | Claims system | Handles disputes |
| Policy-service changes (add driver, vehicle) | With approval | AMS | Approves rating changes |
| Certificate-of-insurance requests | Fully | AMS | None |
| Renewal and cross-sell outbound calls | With approval | AMS, dialer | Approves the script and consent list |
| Binding coverage or giving advice on what to buy | No | AMS | Always a licensed agent |
Off-the-shelf AI agents on the market
Real products you can sign up for today — with current 2026 pricing, honest pros and cons.
Sonant
P&C agencies on one of its supported AMS platforms
Demo only
Contact sales; no public pricing
Applied Certified Vendor for Applied Epic since August 2026
Pros
- +Purpose-built for P&C agency quote intake, FNOL and call routing
- +Integrates with Applied Epic, EZLynx, HawkSoft and AMS360
- +Certified status with Applied Epic signals a validated integration, not a workaround
Cons
- −No public pricing
- −Narrower focus than a general-purpose customer-service vendor
Liberate
Carriers and larger agencies wanting both inbound service and outbound sales from one voice agent
Demo only
Contact sales; no public pricing
Raised a $50M Series B in October 2025 at roughly a $300M valuation
Pros
- +Voice agent Nicole handles both inbound and outbound sales and service
- +Vendor-reported results include a 263% ROI case and 14x claim-volume handling during a hurricane surge
- +Backed by Battery Ventures with real enterprise traction
Cons
- −No public pricing
- −Vendor-reported ROI and volume figures aren't independently audited
Salesforce Agentforce (Financial Services Cloud)
Agencies and carriers already running Salesforce Financial Services Cloud
Salesforce Foundations starting point
$2 per conversation, or Flex Credits at $500 per 100,000 credits (a standard action costs 20 credits, a voice action 30)
Agentforce 1 / Max edition from $550/user/mo, including 2.75M Flex Credits per org per year
Pros
- +Published, predictable per-conversation and credit pricing, unlike the insurance-specific vendors
- +Works on top of Financial Services Cloud, which many larger agencies and carriers already run
- +Backed by Salesforce's broader agent ecosystem and roadmap
Cons
- −Requires the Salesforce and Financial Services Cloud ecosystem
- −Not insurance-specific out of the box; quote intake and FNOL workflows need configuration
The AI stack
A P&C insurance agent needs a reasoning model, access to the agency management system, real tool access to quote and check claims, and a hard boundary around what only a licensed agent can say.
Reasoning model
Drives quote intake, FNOL capture and the tone-sensitive work of outbound renewal calls
Claude Haiku 5.5
$0.10/M input tokens (up to 100K), $0.50/M outputRouting and classification
Claude Sonnet 5.5
$2/M input, $10/M outputThe main quote and FNOL intake loop
Claude Opus 5.5
$4/M input, $20/M outputOutbound renewal and cross-sell calling
Our pick: Claude Sonnet 5.5 or GPT-6.1 Sol for quote intake and FNOL, with Claude Opus 5.5 for outbound renewal calls where tone and timing matter more.
Knowledge & retrieval
Grounds quote, policy and claims answers in your actual AMS data and carrier guidelines
AMS-native data (Applied Epic, EZLynx, HawkSoft)
Included in your AMS subscriptionQuote and claims lookups
Custom RAG over carrier appetite and underwriting guidelines
Roughly $0.15/mo on Supabase pgvector for a small corpusCustom builds
Our pick: Pull policy and claims data straight from your AMS; add custom RAG only once you need to blend carrier appetite guidelines the AMS itself doesn't expose.
Integrations / tools
Lets the agent actually look up a policy, log a claim or update a record
Native AMS connectors (Sonant, Liberate)
Included on vendor plansBuy-SaaS
Custom AMS API integration
Developer timeCustom builds and n8n
Our pick: Use a certified AMS connector wherever your agency management system supports one, and build a custom API integration only when it doesn't.
Observability & evals
Catches a licensed-activity or consent mistake before it reaches more callers
Vendor dashboards
IncludedBuy-SaaS
Langfuse with a licensed-activity review
Free up to 50K units/mo, then $29+/moCustom builds
Our pick: Have a licensed agent review a sample of transcripts weekly specifically for anything that reads as unlicensed advice or a binding commitment.
Reference architecture
The hardest engineering problem is the licensed-activity boundary: the agent has to move a quote or claim forward convincingly without ever binding coverage or advising on what to buy, both of which require a license in most states, so the handoff point has to be conservative and consistent.
Call or request arrives
Phone line, website, or claims intakeA quote request, FNOL or policy-change call lands with the agency or carrier
Identity and policy lookup
Agency management systemThe agent verifies the caller and pulls the relevant policy, quote or claim record
Capture and classify
Reasoning modelThe agent captures intake data (vehicle, driver, loss details) and classifies what action is needed
Licensed-activity check
Guardrail layerAnything that would bind coverage, advise on a purchase, or needs adjuster judgment routes to a licensed human
Action or routing
AMS, claims systemApproved administrative actions, a certificate request, a status update, execute directly; everything else routes
Human handoff
Agent or adjuster queueRouted requests carry full intake context so staff don't re-ask what the caller already said
Logging and eval
Observability toolEvery call is logged, and a licensed agent reviews a sample weekly for licensed-activity drift
Estimated cost per request
~$0.03-$0.08 per handled call in model costs alone, before any vendor fee
Cost calculator
Drag the sliders to model your actual usage. The numbers update in real time so you can stress-test economics before writing a single line of code.
Estimate your monthly cost at your own agency's call volume and outbound mix.
Estimated monthly cost
$140
≈ $1,680 per year
Calculator notes
- Sonant and Liberate publish no per-call rate, so this calculator models a custom build or DIY stack; get a written quote from either vendor before comparing against your own volume.
- The outbound cost line assumes TCPA-compliant consent logging and DNC scrubbing on every outbound call, which adds real process cost beyond telephony minutes.
- Model cost applies on every path, since the underlying reasoning calls don't disappear just because you bought a SaaS license.
- Add your existing AMS subscription fee on every path; none of this replaces Applied Epic, EZLynx or HawkSoft.
Build it yourself with vibe-coding tools
n8n can run a basic quote-intake or FNOL agent against your AMS's API, though licensed-activity review has to happen before it goes live, not after.
Time to MVP
1-3 days for a first version, 2-4 weeks to production quality with licensed-activity and TCPA review
Total cost to MVP
$0-$70 (n8n Cloud Starter plus model API costs)
You'll need
Starter prompt
Set up a webhook trigger on new quote requests or FNOL calls from your phone system or website form. Feed the request into an AI Agent node using Claude Sonnet 5.5, with an HTTP Request tool for your agency management system's API to look up or create a policy, quote or claim record. Give the AI Agent node this system prompt: 'You are an intake assistant for {Agency Name}, a licensed P&C insurance agency. You capture quote and first-notice-of-loss information accurately using the AMS tool, and you confirm details back to the caller. You never bind coverage, quote a final premium, or advise the caller on what coverage to buy: those require a licensed agent. If the caller asks for advice on coverage, or the loss involves injury or a large claim, reply with ESCALATE and a one-sentence summary for a licensed agent to call back. State that you are an AI assistant.' Add a Switch node that routes ESCALATE requests to a licensed agent's queue immediately and logs every intake to your AMS record.
Paste this into n8n
Follow-up prompts (run in order)
- 1
Add a TCPA consent check before any outbound renewal call
- 2
Add a daily summary of quotes and claims captured versus escalated
- 3
Add a certificate-of-insurance self-serve flow for existing policyholders
- 4
Add a catastrophe-surge mode that widens after-hours coverage automatically
Expected output
A working agent that captures quote and FNOL details accurately and hands anything requiring a license to a human with full context attached.
Known gotchas
- !Licensed-activity rules on what an AI can say vary by state, so a script that's fine in one state may not be in another
- !The AI Agent node will confidently quote a premium or give coverage advice from general training data unless explicitly instructed not to
- !Webhook retries can double-fire on some phone systems, so dedupe on the call or claim ID
- !There's no built-in TCPA consent logging, so you have to build and audit it yourself
- !AMS API completeness varies significantly between Applied Epic, EZLynx, HawkSoft and AMS360
Compliance & risk reality check
A P&C insurance agent sits close to two separate regulatory lines: what only a licensed person can say, and how it contacts people by phone.
Licensed-activity limits
In most states, only a licensed agent can bind coverage or advise a customer on what to buy, and AI agents must stay inside intake, routing and status-update tasks; specifics vary by state and should be confirmed with counsel.
Mitigation: Have a licensed agent define and sign off on exactly what the AI can say, and escalate anything resembling advice or binding authority.
TCPA for outbound lead and renewal calls
Outbound AI voice calls to leads or policyholders need prior express consent under the FCC's 2024 ruling that AI voices count as artificial voices under the TCPA; the one-to-one consent rule for shared leads was vacated in January 2025, but ordinary consent requirements still apply.
Mitigation: Capture and log written consent naming AI and automated calls, scrub against DNC lists, and disclose the AI at the start of every call.
GLBA safeguards
Non-bank financial institutions, including many insurance agencies, must maintain a written information security program and notify the FTC within 30 days of a breach affecting 500+ consumers' unencrypted data.
Mitigation: Apply encryption, access controls and vendor oversight to any system handling policyholder data, including the AI agent's logs.
Vendor training on policyholder data
AI vendors process policyholder conversations to generate responses, and some contracts allow using that data to improve their own models.
Mitigation: Add a no-training clause to vendor contracts where offered.
Build vs buy: the real math
3-6 weeks for a RapidDev build; 1-3 days for a basic n8n version
Custom build time
$13K-$25K for an agency build (up to $35K for multi-AMS work); Sonant and Liberate are quote-only
One-time investment
Hard to calculate against Sonant or Liberate since neither publishes pricing; the practical comparison is time to a working, licensed-reviewed system
Breakeven vs buying
Because Sonant and Liberate both price through a sales conversation, there's no public rate to benchmark a custom build against the way there is in customer service. What's comparable is speed and control: a RapidDev build at $13K-$25K can be live in 3-6 weeks with your own licensed staff defining the script, while a vendor sales process for an insurance-specific AI product routinely takes weeks before you even see a number. For an agency already deep into one AMS with a certified vendor option, getting that vendor's actual quote first is still worth doing before committing to a custom build.
Skip the DIY — RapidDev builds the production version
A Lovable MVP gets you a demo. Production needs auth that doesn't leak data, AI calls that don't bankrupt you, observability when models drift, and code you can audit. That's what we ship through our AI app development service.
Discovery call (free)
30 minWe map your exact P&C insurance AI agent use case: who uses it, target volume, AI model choice, integrations, compliance scope. You get a detailed scope document and fixed-price quote within 48 hours.
AI-accelerated build
3-6 weeks for a RapidDev build; 1-3 days for a basic n8n versionOur engineers use Claude Code, Lovable, and custom tooling to ship 3–5x faster than agencies. You see weekly progress in a staging environment — not a black box.
Launch + handoff
1 weekWe deploy to your infrastructure, transfer the GitHub repo, set up CI/CD and monitoring, and train your team. You own 100% of the source code, prompts, and model configurations.
What you get
Timeline
3-6 weeks for a RapidDev build; 1-3 days for a basic n8n version
Investment
$13K-$25K for an agency build (up to $35K for multi-AMS work); Sonant and Liberate are quote-only
vs SaaS
ROI in Hard to calculate against Sonant or Liberate since neither publishes pricing; the practical comparison is time to a working, licensed-reviewed system
30-min call. Fixed-price quote within 48 hours. No commitment.
Frequently asked questions
How much does an AI agent for P&C insurance cost?
Sonant and Liberate both price through a sales conversation with no published rate. Salesforce Agentforce charges $2 per conversation or Flex Credits at $500 per 100,000 credits. A custom RapidDev build runs $13K-$25K upfront (up to $35K for multi-AMS work), plus $150-$600/mo in model and infra costs. A DIY n8n version costs $0-$70/mo.
How long does it take to set up?
Sonant and Liberate's sales and onboarding process typically takes several weeks before go-live. A custom agency build takes 3-6 weeks because it's wired into your exact AMS and licensed-activity rules. A basic n8n version can work within 1-3 days, though a proper licensed-activity and TCPA review takes 2-4 weeks.
Can RapidDev build this for us?
Yes. RapidDev has shipped 600+ apps and offers a free 30-minute consultation to scope your AMS, call volume and state licensing requirements before you commit to anything.
Can an AI agent actually bind coverage?
No. In most states, binding coverage and advising a customer on what to buy require a license, so a well-built agent handles intake, routing, status updates and certificate requests, and hands off anything involving binding authority or coverage advice to a licensed agent.
What happens during a catastrophe surge?
Vendors like Liberate market this as a key use case, reporting one carrier handled 14 times its normal claim volume during a hurricane without adding staff. That figure is vendor-reported, not independently audited, but after-hours and surge coverage is the clearest wedge for AI agents in this industry.
Is it legal to call insurance leads with an AI voice?
Yes, with consent. The FCC's 2024 ruling requires prior express consent for AI-generated voice calls, and while the one-to-one consent rule for shared leads was vacated in 2025, ordinary TCPA consent and disclosure rules still apply to every outbound call.
Can it work with Applied Epic, EZLynx or HawkSoft?
Yes. Sonant became an Applied Certified Vendor for Applied Epic in August 2026 and also integrates with EZLynx, HawkSoft and AMS360. A custom build can connect to any of these through the AMS's own API.
Want the production version?
- Delivered in 3-6 weeks for a RapidDev build; 1-3 days for a basic n8n version
- You own 100% of the code
- AI cost monitoring built in
30-min call. No commitment.
