Insurance forms are among the longest, most complex web forms on the internet. A typical online claims form asks for policy numbers, dates of incidents, vehicle identification numbers, descriptions of damage, contact details for all parties involved, and more. Many of them are never finished: across the insurance forms in Zuko’s industry table, 55.81% of 5.3 million tracked sessions converted, and 47.80% of mobile sessions (Zuko benchmark, 2025).

That abandoned form is not just a UX problem. It is a delayed claim, a frustrated policyholder, and a phone call to your call center — a cost you can read off your own call center figures. Multiply that across thousands of claims per month, and the business impact is staggering.

Voice form filling offers a direct solution: let policyholders speak their claim details naturally, and let AI fill every field at once. No typing, no tabbing between fields, no squinting at labels on a mobile screen. This article explains how it works, why insurance forms specifically benefit, and what it takes to implement.

Why Online Claims Forms Have the Worst Abandonment Rates

Not all form abandonment is equal. A newsletter signup form with two fields has a very different abandonment profile than an insurance claims form with 25 fields spread across three pages. Insurance forms are uniquely punishing for several reasons — and multi-step layouts make it worse. Many insurance portals break claims forms across several pages, and every page transition is one more point where a policyholder can drop off — the losses compound.

Sheer Volume of Fields

A first notice of loss (FNOL) form for auto insurance typically requires: policyholder name, policy number, date of incident, time of incident, location of incident, description of what happened, vehicle year/make/model, VIN, license plate number, other driver's name, other driver's insurance carrier, other driver's policy number, police report number, witness information, injury details, and estimated damage amount. That is a long form before any optional section, and many forms ask for even more.

Each field is a decision point where the user can give up. Research consistently shows that form completion rates drop with every additional field. By the time you reach 20+ fields, the majority of users who started will not finish — especially on mobile.

Data Users Do Not Have Memorized

Unlike a contact form where users know their own name and email, insurance forms require information people do not carry in their heads. Policy numbers, VIN numbers, claim reference numbers, and exact dates of incidents require looking things up. Every time a user has to leave the form to find a piece of information, the chance of abandonment spikes.

Mobile Users Reporting from the Scene

A significant percentage of auto claims are filed from the scene of an accident — on a phone, standing on the side of the road, possibly shaken up. Typing 25 fields on a 6-inch screen with a virtual keyboard is not just inconvenient; it is effectively asking someone in a stressful situation to do precision data entry. Many give up and call the claims hotline instead, costing the insurer far more per interaction.

Elderly and Less Tech-Savvy Customers

Insurance companies serve a disproportionately broad demographic. Home insurance, life insurance, and Medicare supplement policyholders skew older. Many of these customers find small form fields, dropdown menus, and date pickers genuinely difficult to use. They are not abandoning because they do not want to file — they are abandoning because the interface defeats them.

How Voice Form Filling Works for Insurance

Voice form filling replaces the field-by-field typing process with a single spoken statement. The policyholder clicks a Fill by Voice button on the form and speaks naturally — just as they would describe the situation to a claims agent on the phone.

"My name is John Smith, policy number AB-123456, the incident happened on March 10, 2026, it was a rear-end collision at the intersection of Main Street and Oak Avenue, the other driver was Sarah Johnson, estimated damage is about $3,500."

The AI processes this single utterance and fills the corresponding fields simultaneously: Name → John Smith, Policy Number → AB-123456, Date of Incident → 03/10/2026, Incident Type → Rear-end collision (selected from dropdown), Location → Main Street and Oak Avenue, Other Driver Name → Sarah Johnson, Estimated Damage → $3,500.

Seven fields filled from one sentence. No tabbing, no clicking into each field, no formatting dates or selecting from dropdowns manually. The user reviews the filled data, corrects anything that needs adjustment, and submits.

Handling Complex Field Types

Insurance forms are not simple text inputs. They include dropdowns (incident type, state, vehicle make), date pickers, currency fields, phone number formats, and radio buttons. A voice form filling AI that only handles text inputs would miss half the form.

TypelessForm handles all standard HTML form elements: it selects the correct dropdown option when a user says "rear-end collision," formats dates correctly regardless of how the user says them ("March tenth" or "3/10" or "the tenth of March"), and enters currency amounts with proper formatting. This matters for insurance because the form field diversity is unusually high. For a deeper look at how AI handles different field types, see our comparison of AI form filling tools.

Works Where It Matters Most: Mobile

Voice input is inherently a mobile-first interaction. Speaking into a phone is natural — typing 25 fields into a phone is not. For policyholders reporting auto claims from an accident scene, or homeowners filing damage claims during a weather event, voice form filling turns typing 25 fields on a phone into one spoken account of what happened, then a review.

No app download is required. The voice widget works directly in the mobile browser — Safari on iOS, Chrome on Android — because it is a web-based JavaScript widget, not a native app.

Use Cases in Insurance

Voice form filling applies across the insurance value chain, not just claims. Here are the most impactful use cases.

Claims Filing (Auto, Home, Life)

This is the highest-impact use case. First notice of loss forms are long, stressful, and time-sensitive. Auto claims benefit from mobile voice input at the scene. Home insurance claims benefit when policyholders can describe damage naturally rather than categorizing it into form fields. Life insurance claims, filed during difficult personal circumstances, benefit from reducing the mechanical burden of data entry, except for sections that ask for diagnoses or other health information: those are outside what TypelessForm is built for.

Quote Requests

Insurance quote forms ask for extensive personal and property information: age, address, vehicle details, coverage history, number of drivers in household. These forms are the top of your sales funnel, and every abandonment is a lost potential customer. Voice input makes the quote process feel conversational rather than bureaucratic — a customer can say "I'm 34, married, live at 456 Elm Street, Apartment 2B, Austin Texas, drive a 2022 Honda Civic, clean driving record" and get most of the form filled in one go.

Policy Applications

New policy applications often require even more data than claims forms: employment information, property details (for home insurance), and beneficiary information. These are the forms that most benefit from voice filling because they combine high field counts with information the applicant knows but would normally have to type out character by character.

Customer Onboarding and Renewal Forms

When a customer switches carriers, they need to provide all their existing policy details, coverage preferences, and personal information. Making onboarding forms voice-enabled reduces friction at the exact moment when the customer has decided to give you their business. The same applies to annual renewals, where forms pre-fill some data but still require updates — voice input lets customers state in one sentence what has changed ("I moved to 789 Pine Road, and I added a 2025 Toyota Camry") without navigating through unchanged fields.

Hotels face similar booking form challenges — see our guide on voice form filling for hotel booking.

Security and Compliance for Insurance Data

Insurance IT teams evaluate every vendor through a compliance lens — and rightly so. Claims data includes some of the most sensitive consumer information: policyholder PII (names, addresses, dates of birth), policy numbers, Social Security numbers on certain forms, medical information for health and life claims, and financial details like damage estimates tied to specific individuals. State insurance commissions impose data handling requirements that vary by jurisdiction, and carriers operating across multiple states must satisfy the strictest among them. Forms that collect diagnoses or other protected health information (PHI) are outside what TypelessForm is built for: HIPAA is out of scope, and those fields should be typed and marked data-ai-private.

Any technology that touches this data pipeline needs to pass your InfoSec review. Here is how voice form filling holds up.

No Audio Kept on Our Infrastructure

TypelessForm processes voice input in memory and extracts structured text values (the field data) for the form. We keep no voice recordings on our own infrastructure: audio is discarded as soon as the response returns. OpenAI, our speech provider, does not keep the audio. For its own abuse monitoring it keeps the requests and responses of its language-model calls for up to 30 days: the transcript text, the descriptions of the form fields and the filled values. We cannot have it deleted sooner. It does not use this data to train its models. For your compliance team: account for OpenAI’s 30-day retention of the requests and responses of its language-model calls (transcript, field descriptions and filled values) when reviewing the processing flow.

What Stays in Your Claims Pipeline and What TypelessForm Handles

TypelessForm fills HTML form fields with text values — identical to what a policyholder would have typed manually. Your existing form submission logic, encryption (TLS in transit, encryption at rest), validation rules, and backend claims processing remain completely unchanged. The widget does not intercept form submissions and does not communicate with your claims management system API. What does change is who handles the voice: the voice and the structure of the form go through our API to OpenAI, so Webappski, which operates TypelessForm, processes that data on behalf of the site operator, with OpenAI as a sub-processor. The values already in the form's fields, unless a field is marked data-ai-private, go to our API with the request that fills the form; we do not store them and do not pass them to OpenAI. For your vendor review, TypelessForm and OpenAI are the providers to assess, as with any other provider in your claims flow. The security page lists the sub-processors and regions, links the Data Processing Agreement and says how to request the signed copy and our security questionnaire.

Sensitive Fields Are Not Filled by Voice

Insurance forms frequently sit adjacent to payment fields, Social Security number inputs, and authentication fields. TypelessForm automatically detects these sensitive field types and excludes them from voice filling: SSN fields, credit card numbers, CVV codes, passwords, and OTP inputs. Even if a policyholder speaks a Social Security number aloud, the system will not fill it into an SSN field. The spoken digits are still part of the audio sent to OpenAI for transcription and of the transcript OpenAI holds for up to 30 days, so applicants should type a Social Security number, not say it. The value of any field of the same form that is not marked data-ai-private, the Social Security number field itself included, goes to our API with the request that fills the form, however it got there (typed, pasted or filled in by the browser); we do not store it and do not pass it to OpenAI, and a field marked data-ai-private is cut out of that request, so mark SSN fields data-ai-private. This matters particularly for claims forms that collect both general incident data and sensitive financial or identity information on the same page. For details on how this exclusion works across different tools, see our guide on automatic form filling methods.

Regulatory Alignment

TypelessForm is fully GDPR compliant — no voice recordings kept on our infrastructure (OpenAI, the speech provider, does not keep the audio, but for abuse monitoring it keeps the requests and responses of its language-model calls (the transcript, the form-field descriptions and the filled values) for up to 30 days), no user profiles created, no biometric identifiers derived. For US insurance carriers, what matters is the data path, so here it is. The widget script runs in the policyholder's browser, and the voice goes from there to our API (a Google Cloud Function in the EU, europe-central2), which sends the audio to OpenAI in the United States for speech-to-text and the resulting text to an OpenAI language model that maps it to your form fields. The values already in the form's fields, the Social Security number field included unless it is marked data-ai-private, go to our API with the request that fills the form; they are not passed to OpenAI. TypelessForm stores no recordings, no transcripts and no filled values, and its logs keep only metadata. The filled values are returned to your page, and from there your own systems store and process them under your own controls. TypelessForm is not SOC 2 certified, so assess this data path against your own SOC 2 and NAIC Insurance Data Security Model Law controls; the sub-processors, regions and retention are listed on our security page.

Business Impact

The business case for voice form filling in insurance rests on four measurable outcomes.

Reduced Form Abandonment

The primary metric. When a 25-field form can be substantially filled with two or three spoken sentences instead of 25 individual typing actions, the completion rate improves. The effect is most pronounced on mobile, where typing friction is highest, and with older users, where interface complexity is the primary barrier. On a high-volume claims form, even a small lift in completion adds up over a year of claims — measure yours with a pilot on one form against your historical baseline.

Claims That Stay Digital

Channel matters in claims. A policyholder who can speak the claim instead of typing every field on a phone is more likely to file digitally rather than calling the claims center. Every claim that stays in the digital channel is a routine call your call center does not take. At scale, this shifts meaningful volume from expensive phone channels to low-cost digital self-service.

Improved Mobile Experience

Field adjusters inspecting damage on-site, policyholders at accident scenes, and homeowners documenting storm damage all need to interact with forms on mobile devices. Voice input turns these forms from a mobile liability into a mobile advantage. A field adjuster can speak property details, damage assessments, and notes directly into the inspection form without juggling a clipboard and a phone keyboard.

Accessibility Compliance

Insurance companies serve a broad population that includes elderly customers, people with motor disabilities, and people with visual impairments. The Americans with Disabilities Act (ADA) and the European Accessibility Act (EAA), which takes full effect in June 2025, require digital services to be accessible. Voice input is one of the most practical ways to improve form accessibility — it provides an alternative input method for users who cannot effectively use keyboards or touchscreens. For a comprehensive look at how voice input supports WCAG compliance, read our voice form filling accessibility guide. For a comprehensive overview of accessibility tools, see our best form filling tools for accessibility guide.

Integration with Claims Portals and Insurance Platforms

Insurance IT teams rightly worry about integration complexity. Your claims portal is not a simple marketing website — it is typically built on top of a claims management system like Guidewire ClaimCenter, Duck Creek Claims, or Majesco Policy, with custom front-end layers, role-based access, and specific data flow requirements. Adding a new technology layer to that stack usually means months of vendor evaluation, API integration, and QA cycles.

TypelessForm skips all of that. Here is why.

One Script Tag on Your Portal Page

The entire integration is a single JavaScript tag added to the HTML of your claims portal page — the same page where policyholders fill out the FNOL form, the quote request, or the policy application. The widget automatically detects form fields on the page and adds the Fill by Voice button. There is no build step, no npm package to maintain, no framework dependency, and no server-side deployment.

Works at the HTML Form Level, on Portals That Render Standard Fields

Because TypelessForm operates at the HTML form level — interacting with standard DOM elements like inputs, selects, and textareas — it can fill the standard form fields of any claims portal that renders them in a browser, whether the portal sits on a claims platform or was built in-house. We have not run it separately on Guidewire, Duck Creek or Majesco portals, so test it on your own portal before you rely on it. If a policyholder can type into a standard field on your portal, the widget can usually fill it by voice, except the sensitive fields described above.

Claims Portal, Self-Service Portal, Agent-Facing Forms

Voice form filling is not limited to the policyholder-facing FNOL form. It should work equally well on agent-facing intake forms — where a claims representative is entering details from a phone call and can speak them instead of typing — and on self-service portals where policyholders update their address, add vehicles, or request policy changes. A form of your portal that the page renders with standard fields should be reachable by adding the script tag to that page; a form shown in an iframe or on another domain is not.

No Backend Changes, No API Integration

On your side, TypelessForm is a script tag on the page. It fills form fields with text values — the same values a user would have typed. Your form's existing validation rules, submission logic, API calls to Guidewire or Duck Creek, and backend claims processing pipelines remain completely unchanged. There is no webhook to configure, no middleware to deploy, and no changes to your claims data model.

Three-Step Deployment

From adding the script tag to having voice input live on a claims form takes three steps. For insurance IT teams accustomed to six-month integration timelines for new vendor tools, this is a fundamentally different deployment model. You can try the live demo without any signup to see how it works on a sample form, then deploy to a single claims form page for a pilot before rolling out more broadly.

Getting Started: Start with Your FNOL Form

The fastest way to see results is to start with your highest-abandonment form — and for most carriers, that is the First Notice of Loss (FNOL) claims form. It is the longest, the most stressful for policyholders, the most likely to be filled on mobile, and the most expensive when it gets abandoned (because every incomplete FNOL becomes a phone call to your claims center).

Here is the recommended pilot approach:

  1. Identify your baseline — pull your current FNOL form completion rate and average time-to-complete from your analytics. If you do not track these yet, set up basic event tracking before the pilot.
  2. Add the widget to one form — deploy TypelessForm on your FNOL page with a single script tag. Three steps from start to live: get a key, add the script tag, set the key.
  3. Run for 30 days — measure completion rate, average fill time, and the percentage of users who choose voice versus typing. Compare these numbers against your baseline.
  4. Calculate the impact — every additional completed digital FNOL is a phone call that did not happen. Multiply the extra completed digital claims by your own cost per inbound claims call. At 10,000 FNOL starts a month, every percentage point of completion is 100 more digital claims a month.
  5. Expand — once you have measured the impact on FNOL, roll out to quote request forms, policy applications, and customer onboarding forms.

TypelessForm offers a free pilot with 200 form fills — enough to test on a real FNOL form with real policyholders and measure the impact before any commitment. Paid plans start at €29 per month (checked 3 October 2026) and scale with volume.

Try the live demo to see how voice form filling works on a sample form — no signup required.

Insurance forms are hard. They ask for a lot of information, from users who are often stressed, on devices that make typing painful. Voice form filling does not make the form shorter — it makes filling it painless. And in an industry where every completed digital FNOL is money saved and a customer retained, that difference matters.

Revised 4 and 5 October 2026: the voice-storage lines in this article now match our security page. We keep no voice recordings on our own infrastructure. OpenAI, our speech provider, does not keep the audio, but for its own abuse monitoring it keeps the requests and responses of its language-model calls (the transcript text, the descriptions of the form fields and the filled values) for up to 30 days, and we cannot have it deleted sooner. Earlier wording about what is held, and for how long, is replaced. The lines on sensitive fields now say that such fields are not filled by voice; digits a person says aloud still reach transcription.

Revised 5 October 2026: the sections Regulatory Alignment, "What Stays in Your Claims Pipeline and What TypelessForm Handles" (retitled), "No Audio Kept on Our Infrastructure" (retitled), "Sensitive Fields Are Not Filled by Voice" (retitled), "No Backend Changes, No API Integration", the section on claims platforms (retitled) and its FAQ answer, and the FAQ answer on security now describe the data path and what we have checked: the voice and the structure of the form go through our API to OpenAI, Webappski, which operates TypelessForm, processes them on behalf of the site operator with OpenAI as a sub-processor, TypelessForm is not SOC 2 certified, HIPAA is out of scope and forms that collect diagnoses or other health information are outside what it is built for, digits spoken aloud are not removed from the audio so a Social Security number should be typed, and we have not run it separately on Guidewire, Duck Creek or Majesco portals. Earlier wording that described TypelessForm as a tool that adds nothing to your data flow, or as tested on those platforms, is replaced. The section on sensitive fields is retitled "Sensitive Fields Are Not Filled by Voice" and now says that spoken digits are part of the audio sent to OpenAI for transcription and of the transcript OpenAI holds for up to 30 days, and that the value of any field of the same form that is not marked data-ai-private, the Social Security number field included, goes to our API with the request that fills the form, however it got there. The passages on the compliance lens and on the data path say the same about the values already in the form's fields, and the section on portals is retitled "Works at the HTML Form Level, on Portals That Render Standard Fields": it no longer says that the widget can fill any field of any portal.

Revised 5 October 2026 (second pass): the line on agent-facing intake forms and self-service portals now says that voice form filling should work equally well there, because we have not run it separately on Guidewire, Duck Creek or Majesco.