Automate underwriting
Unstract reads each broker submission as it arrives, from the ACORD applications to the loss runs and the schedule of values, clears it, checks it against your appetite, and sends the clean ones on to your AMS, rater or policy system.
Trusted by insurance teams around the globe
Submissions pile up faster than underwriters can read them
Insurance underwriting starts with a broker's email full of attachments. An underwriting assistant opens each one, keys the applications, retypes the schedule of values and reads five years of loss runs before an underwriter can say whether the risk is even a fit.
An ACORD 125 keyed by hand while the broker's quote date gets closer
A loss run scanned crooked and valued five months ago
A loss run older than your guidelines allow, found after the quote work began
The same risk sent by two brokers and worked twice
An out-of-appetite class spotted only after the policy was bound
Every submission waits in a queue to be keyed and screened, and the broker places the risk elsewhere while it waits.
Unstract reads the whole submission, clears it, checks it against your appetite and authority, and sends the clean ones straight to rating. Underwriters see only the submissions with a named gap.
From the broker's email to the rater and workbench, without the keying
Five steps, run on their own for every submission. Parsing by LLMWhisperer, extraction by Document Studio, the whole run by Flow.
Pick up broker submissions wherever they land
Brokers send ACORD 125, 126 and 140 applications, loss runs, schedules of values and supplemental questionnaires by email, and some of it follows later through the portal. Every attachment is picked up where it arrives and filed under one submission, so the supplemental that turns up an hour late joins the right record.
Nobody sorts the underwriting inbox by hand. The submission exists the moment the broker's email does.
Clear the insured before anyone works the risk
The named insured, FEIN and mailing address are matched against open submissions and in-force policies. A second broker sending the same machine shop is caught here, and so is an account you already write. Earlier quotes on the same FEIN are shown to the underwriter with their outcome.
Two assistants never quote the same risk for two brokers.
Read every application, then pull the fields that matter
LLMWhisperer reads each document with its layout kept, so a faxed ACORD 125, a scanned loss run and a schedule of values in a spreadsheet all come out readable. Document Studio then extracts the fields you defined: named insured, FEIN, class codes, revenue and payroll, each location with its building and contents values, and five years of losses with dates, paid and reserved amounts.
Every broker's version of the schedule of values lands in the same fields.
Check it against your appetite before anyone touches it
Class, territory, limits and loss history are checked against your appetite guide and each underwriter's authority, and the documents against what the line of business requires. A missing questionnaire or a loss run older than 90 days holds the submission in a review screen with the rule, the document and the value side by side.
Write the referral rules once (out of appetite, above authority, losses over threshold, a low-confidence read) and only those submissions wait.
Send clean submissions to rating where underwriters already work
A clean, in-appetite submission goes straight through to rating with nobody keying it. One with a gap stops for an underwriter or an underwriting assistant with the gap named. Either way the extracted data lands in your rating tool, your underwriting workbench and your policy administration system, Guidewire PolicyCenter, Duck Creek Policy or another.
Underwriters open a submission that is already keyed, cleared and screened.
Watch a week of your submissions go through it.
Bring the broker emails from one line of business, loss runs and odd schedules included. We run them live and show each field, each appetite rule and each referral.
Everything a broker sends with a submission
Documents
- ACORD forms
- Loss runs
- Statements of values
- Supplemental questionnaires
- Life insurance applications
- Annuity applications
- Cargo insurance forms
- Certificates of insurance
- Loss and damage claims
- Financial statements
- Inspection and loss control reports
- Vehicle and driver schedules
- Attending physician statements
- Prior policy declarations
- Broker cover emails and risk narratives
- Open submissions and in-force policies from your policy system
Fields captured
Our agents are extroverts.
Cloud storage like S3, Google Drive, Dropbox and Box. Databases and warehouses like Snowflake, BigQuery, Postgres and SQL Server. Accounting, ERP and CRM systems like QuickBooks, Xero, SAP, NetSuite, Salesforce and HubSpot. Project trackers like Jira, Asana, ClickUp and Linear. And anything with an API.
Start with one line of business
Your team keeps its rater and workbench, and nothing is switched over at once. The run is built on your own submissions and proven on one line before the next one joins.
Book a consultation
We go through a week of submissions with you: what brokers send, how clearance works today, and which appetite rules decide a referral.
Built on your submissions
The fields, your appetite and authority rules, the clearance against your policy data and the hand-off to rating are set up on real broker submissions from your inbox.
Pilot one line, then add the rest
The first line is live within days. When the referral rate and the turnaround hold, the next line joins with its own appetite guide.
Built for every desk a submission crosses
Client applications and loss runs read once and sent to market complete, so carriers stop coming back for the missing supplemental.
Submissions triaged against each carrier's appetite and your binding authority before an underwriter opens the file.
Clean, in-appetite submissions sent straight to rating, and referrals that arrive with the gap already named.
Additional insured and waiver requests captured from the submission, so certificates issue from the bound policy without anyone retyping them.
Submissions screened in minutes, quotes back to brokers sooner
Built to hold an insured's financials
A submission carries an insured's FEIN, revenue, payroll, property values and claim history, and life applications add medical history. Brokers send it on trust: a leak exposes their client and their book.
- Encrypted in transit and at rest.
- Submissions are never used to train models.
- Access by role, line of business and authority level, with every referral decision and correction recorded against the submission.
- Retention you set: processed submissions are deleted on the schedule you choose.
- Self-hosted deployment for carriers and MGAs that keep applicant and medical data inside their own environment.
What underwriting leaders ask first
From commercial submissions: named insured, FEIN, class codes, revenue, payroll, locations with building and contents values, construction and protection class, limits requested, and every loss on the loss runs with its date, paid and reserved amounts. From life and annuity applications: the applicant, face amount, riders, beneficiaries and the health answers. You define the fields, and Document Studio returns each one with the page it came from.
Look for software that reads a faxed ACORD form, a scanned loss run and a schedule of values in a spreadsheet with one set of field definitions, and that ties them to one submission. Carriers, MGAs and wholesalers underwriting insurance for commercial accounts get every format from every broker. Unstract reads them all into the same fields and checks them together.
Loss runs, risk narratives and supplementals have no fixed layout, so templates break on them. LLMWhisperer reads each page with its layout kept, and Document Studio extracts the fields by meaning, so a loss run from one carrier and a loss run from another give the same columns. Flow then runs clearance and appetite checks on the result.
Every value Unstract extracts carries a confidence score and a link to the page and position it was read from. Any value below your threshold stops for a person instead of reaching the rater, and every correction and referral decision is recorded against the submission, so an auditor can trace a rated figure back to the broker's document.
Yes. Life insurance applications, annuity applications and attending physician statements are read the same way as commercial submissions: applicant details, face amount, riders, beneficiaries, health answers and replacement questions. Missing signatures or unanswered health questions are flagged before the case reaches an underwriter.
Yes, if you want them to. Straight-through processing sends a submission to rating with nobody touching it when it clears, sits in appetite and authority, has every required document and reads above your confidence threshold. You decide which lines and limits qualify, and everything else goes to an underwriter with the reason attached.
The steps before judgment: submission intake, clearance, reading and keying, and the appetite and completeness screen. Underwriters keep risk selection, pricing decisions and referrals. They start from a submission that is already keyed, cleared and screened.
It stops with the gap named: a missing supplemental, a loss run older than 90 days, a class outside your guide, limits above an underwriter's authority or losses above threshold. The underwriter or assistant requests the document, refers the risk or declines it, the decision is recorded, and the rest of the queue keeps moving.
Your AMS, rater or policy system, depending on who runs the submission. Carriers usually send it to the rating tool, the underwriting workbench and policy administration, such as Guidewire PolicyCenter or Duck Creek Policy. Flow posts it as part of the run.
Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant. Submissions are encrypted in transit and at rest, never used to train models, and kept only as long as you choose. Self-hosted deployment keeps applicant and medical data inside your own environment.
Yes, and most teams do. We build the run on submissions from one line, such as small commercial package or workers' compensation, with your appetite rules and your rater. It goes live there within days, and the next line adds its own appetite guide without rebuilding the run.
Bring a week of broker submissions to the consultation
The emails from one line of business, with the faxed applications, the stale loss runs and the odd schedules of values. You watch each one clear, screen and reach rating, and every referral shows the rule that stopped it.