When a storm hits, claim calls arrive faster than any staffing plan, and the regulatory clocks keep running. VoiceRun agents take first notice of loss, answer claim status and service calls, and qualify buyers, writing structured data into your systems. A coverage opinion, a quote or a bind goes to a licensed person, with the file already built.
Built for VPs of claims, customer service and distribution, and CIOs at carriers, agencies, MGAs and insurance marketplaces.
Commercial claims line
Tue 7:41 AM · CAT event$112.8B
in insured U.S. property losses from natural catastrophes in 2024, up from $80.0B in 2023
Triple-I, Insured Property Losses (Aon data), 202444 days
average time from first notice of loss to final payment on a homeowners claim
J.D. Power, Property Claims Satisfaction Study, 2025777 vs 337
claims satisfaction when customers say their insurer is very easy to communicate with, against very or somewhat difficult
J.D. Power, Property Claims Satisfaction Study, 202529
U.S. jurisdictions have adopted the NAIC model bulletin on insurers’ use of AI, including its section on third-party AI systems
NAIC, AI Model Bulletin adoption map, Oct. 2026The calls
Claims, service and sales calls share one phone line and very different rules. Each of these runs as code against the systems your adjusters, service reps and producers already use.
Takes the loss in structured form: what happened, where, when, injuries, whether the property is safe. Opens the claim, reads back the claim number and next steps, and flags urgent files for triage.
Systems the agent works in: Claims · Policy admin
The highest-volume service call. The agent finds the claim, says where it stands, who owns it and what is still outstanding, and books the next contact when an adjuster is needed.
Systems the agent works in: Claims · Scheduling
When a named event or the end of the business day pushes calls past your staffing, the agent takes the call, tags the claim to the event and books callbacks in priority order.
Systems the agent works in: Claims · Scheduling
Takes the request to add a vehicle, driver, location or lienholder, applies the rules you set for clerical changes, and routes anything that touches coverage to a licensed person.
Systems the agent works in: Policy admin · Agency management
Answers due-date and payment questions, hands payment to your existing payment system, places past-due reminders, and sends reinstatement requests to billing with the file ready.
Systems the agent works in: Billing · Policy admin
Takes certificate requests from contractors and landlords: holder, project, wording needed. Logs standard requests in your agency system and sends special wording, such as additional insured, to the account manager.
Systems the agent works in: Agency management · Policy admin
Qualifies the shopper (line, state, coverage need, timing) and transfers the call to a producer licensed in that state, with context passed along. The quote and the advice happen with the producer.
Systems the agent works in: CRM · Producer roster
Calls a new lead while they are still comparing quotes, after checking consent in code. Qualifies, then transfers to a licensed producer or books a time on a producer’s calendar.
Systems the agent works in: Lead source · Consent record
VoiceScore grades every service and sales call against your scorecard, human or AI, and alerts a supervisor to coverage advice on a service line, a missing disclosure or a consent miss.
Systems the agent works in: Recordings · Scorecard
Why it's hard
Insurance calls carry legal weight. One sentence about coverage from the wrong person, one claim clock that runs out in a storm, or one half-finished loss report is what a lawyer or an examiner finds later.
The reality
In most states, soliciting, negotiating and selling insurance takes a producer license. Unlicensed staff can take information and process clerical changes, but a premium quote or a casual “you’re covered” can become an E&O claim.
How VoiceRun handles it
The line is written in code. The agent captures, records and processes what your rules call clerical. Quotes, coverage questions and binds go to a producer or adjuster licensed in the caller’s state, with the file built. Which topics the agent may handle in which states is configuration your counsel signs off on.
Rules in code
In most states, soliciting, negotiating and selling insurance takes a producer license. Unlicensed staff can take information and process clerical changes, but a premium quote or a casual “you’re covered” can become an E&O claim.
The line is written in code. The agent captures, records and processes what your rules call clerical. Quotes, coverage questions and binds go to a producer or adjuster licensed in the caller’s state, with the file built. Which topics the agent may handle in which states is configuration your counsel signs off on.
Rules in code
Tested before launch
Every line your counsel draws becomes a scripted caller. Before a release reaches a policyholder, simulated callers push on the edges: the coverage question, the premium quote, the caller who is not the insured, the revoked consent.
Simulated callers place real voice calls to the agent, evaluators grade every conversation against what should have happened, and a release that regresses never reaches your callers.
$ vr simulate staging --name licensed-line --wait
7/7 simulated calls passed · evaluator: judge
The platform
Start where the calls hurt most. Agents, QA and the handoff to your licensed people run on one record.
Your systems
The agent reads and writes the records your adjusters, service reps and producers use, so the call ends in the system of record and the next person starts from it. Platforms differ in what they expose, and each call is mapped to the endpoints you have.
Claims system
Opens the claim with structured loss details, reads status, owner and open items, and writes each contact back with a timestamp.
Common Guidewire ClaimCenter, Duck Creek Claims, Majesco
Policy admin & billing
Verifies the policy and the caller’s standing on it, reads coverage dates and billing status, and writes the clerical changes your rules allow.
Common Guidewire PolicyCenter, Duck Creek Policy, Majesco, Sapiens
Agency management
Where an agency’s service work lives: logs claim notices and service requests, submits certificate requests, and puts the note on the account.
Common Applied Epic, Vertafore AMS360, EZLynx, HawkSoft
Producers & leads
Reads which licensed producers can take a transfer in the caller’s state, and the consent on file for a lead before any outbound dial.
Common Salesforce Financial Services Cloud
Telephony
Keeps your numbers and carrier, and hands calls to a producer, an adjuster or your service team.
Common Twilio, Telnyx, Bring your own carrier
Named products are examples of what’s common in each category; none of them is a prebuilt connector. VoiceRun agents call any API or database your team exposes, and every connection is code your team can read. Your carrier, numbers and contact center stay where they are.
Control
Who the agent may talk to, what it may say, when it may dial and who takes the call next are branches in code you own, with the record to show them.
Quotes, coverage questions and binding go to a producer or adjuster licensed in the caller’s state. The routing rule is code your counsel can read, and each handoff is logged with the context it carried.
The AI disclosure and the recorded-line notice are spoken word for word at the top of the call, before anything is captured, in the wording your counsel approves. Recording is a setting you turn on per environment.
Outbound calls read the consent record in code and do not place without it, because an AI voice counts as artificial under the TCPA (FCC 24-17). A revoked consent is honored on the call and written back.
Policy and claim details are returned only after the checks your rules require, and the caller’s role decides what the agent can say. A claimant or repair shop gets the process and an adjuster’s contact.
When a call needs a payment, the agent hands off to your existing payment system, so card numbers stay out of the conversation and the transcript.
Agents are versioned code with release history, regression tests and simulation results. Each call’s transcript, system actions and evaluation sit under one call ID, the documentation a written AI program asks you to hold on a third-party system.
Your code remains your code, your data remains your data, and the models we train for you serve only you.
Deployment options: our SOC 2 managed cloud, your VPC, your cloud, or on-premises with VoiceRun Edge, available under enterprise agreement.
In production
A national insurance marketplace runs VoiceScore on its whole floor.
1M+
call minutes a month
VoiceScore customer
11K
AI evaluations a day
VoiceScore customer
65%
less spent on QA and transcription
VoiceScore customer
FAQ
Our default is no. In most states, soliciting, negotiating and selling insurance takes a producer license, so the agent captures the shopper’s details and transfers to a producer licensed in that state, with the file built. Whether any quoting step counts as clerical in a given state is a question for your counsel, and the agent’s rules are code that follows their answer.
It says a licensed adjuster or producer will answer, then routes the call there with the facts already captured. It never states a coverage position, a fault determination or an outcome, and that behavior is a scenario in the simulation suite that runs before every release. Claim notice, status and process information come from your approved scripts and your systems.
Agents answer in parallel, tag claims to the event and triage on injury and habitability first. The limits to plan for sit around the agent: carrier trunk capacity and the rate limits on your claims system, so we map those with you and rehearse the surge scenarios with simulated callers before storm season. Callers who need an adjuster can take a callback in priority order.
Expect the questions in the NAIC model bulletin on insurers’ use of AI: governance, due diligence on third-party systems, testing, and records an examiner can read. Agents are versioned code with release history, regression tests and simulation results, and each call keeps its transcript, system actions and evaluation. Your counsel decides what the program must say; we supply the evidence.
No. Agents call any API or database your team exposes, the way your own software would. Policy, claims and agency platforms differ in what they expose, so a consultation starts by mapping each call to the endpoints you have. Numbers can be provisioned through VoiceRun or brought from your own Twilio or Telnyx account, and you keep your carrier contracts and rates.
Yes. VoiceScore takes recordings from the platform you run today, by webhook or storage bucket, and grades every call against a scorecard written in plain English, with a quote from the transcript behind each score. Transcripts are redacted before storage and before any model call, and there is no per-seat fee. AI-handled calls land on the same scorecard as your reps.
A consultation is a working session on your call mix, your claims and policy systems, and the licensing line your counsel draws. Bring last storm season’s worst week, and we’ll map which calls an agent takes and which go to a licensed person.
