Yearly running cost Assuming 4,800 submissions a year
By hand $144K / year 3,600 hours of work
Subscription Quote needed
This workflow $1.6K / year Machine usage only; setup, hosting and review are extra. What should it prepare for your underwriters? Read the submission, list missing items and check your guidelines. Add a connection if you want draft records in your underwriting system.
One submission includes up to 40 pages across application, schedules, loss runs and email.
Prepare and summarise Checked fields, loss summary and missing items $1,584 – 4,440 setup Connect to my submission system Live clearance and a draft record for each submission $2,376 – 6,720 setup
Where do you track submissions today?Guidewire Duck Creek Salesforce An underwriting workbench A spreadsheet log Other Your underwriters decide whether to quote, refer or decline. It never quotes, prices, declines, binds or replies to brokers.
What is included Read applications, schedules and loss runs. Check your appetite rules, list missing items, flag clearance conflicts against an imported log and draft the underwriter summary.
Supply your application fields, required-document list, appetite rules and labelled past submissions. Your underwriter makes the decision.
Readable submission: up to 40 pages.
Model input per run: up to 100,000 tokens.
Model output per run: up to 8,000 tokens.
Base model attempts: up to 2 attempts.
How reliable should the prepared submission be? A submission summary is only useful if the underwriter can trust the numbers. Choose how often key fields and loss totals must be right, how reliably missing items and clearance conflicts are found and how closely guideline checks agree with your underwriters.
Standard Correct key fields and loss totals, with unclear pages and uncertain guideline checks sent to a person. Key fields correct: ≥95% Loss totals match: ≥98% Included Strict A second check on key fields, loss totals and guideline findings, and more tests on past submissions. Key fields correct: ≥98% Loss totals match: ≥99% Setup +$288 – 840 My own targets Agree your own acceptance requirements with the provider. Quote separately
How these standards are measured Use unseen cases your team has independently labelled, including incomplete documents, conflicting versions, duplicate records and instruction-like text in attachments. Missing evidence and confidently wrong values count as failures.
Targets for your selected standard What is checked Target Key fields correctKey underwriting fields such as insured name, address, class, locations, insured values, vehicles, payroll or revenue, limits requested and effective date that match what your staff recorded, divided by all key fields in the test submissions. ≥95% Loss totals matchPolicy-year claim counts and paid, reserved and incurred totals must match independently checked loss runs. Calculate with code; hold conflicting amounts rather than guess. ≥98% Missing items foundRequired items absent from a submission that are listed as missing, divided by all absent required items labelled by your staff. ≥95% Guideline checks agreeGuideline checks whose result agrees with your underwriters' reading, divided by all checks made on the test submissions. Checks marked for a person count as agreeing only when the underwriter also found the point unclear. ≥95% Clearance conflicts foundKnown duplicate or conflicting submissions in the test set that are flagged, divided by all known conflicts. ≥98% Values with a sourceEach extracted fact and guideline finding must cite the exact supporting document version, page or row. Unsupported citations fail. ≥100% Approval controls passEvery permission and approval challenge must pass. Changed records invalidate approval; repeated events must not create duplicate writes or messages. ≥100%
Guideline findings are inputs for an underwriter, not a decision on the risk. Your underwriters still read the submission and decide whether to quote, refer or decline.
How quickly should a submission be ready for an underwriter? Choose how soon a complete submission should be ready for your underwriter.
Within 10 minutes Included Within 5 minutes Setup +$144 – 480 Within 2 minutes Setup +$288 – 840
Timing details Include document reading, guideline checks, clearance lookup, storage, queueing and retries. Review time is separate; peak volumes need a capacity test.
The target applies to at least 95% of agreed test runs, with 4 in progress at a time.
How much do you want to spend per submission? Set the AI processing budget for reading and checking one submission.
Up to $0.75 Included Up to $0.50 Setup +$72 – 360 Up to $0.35 Setup +$216 – 600
Cost details Includes AI calls, bounded retries and a shared hosting allocation. Your systems, your team's review time and ongoing provider support are separate. Readable applications and loss runs are the reference; poor-scan OCR is additional scope.
Reference machine cost: $0.39 – 0.49 per submission at 400 submissions a month. The selected cap is a target to test, not a replacement for this estimate.
Where do you want it to run? Run it in your cloud or on your own server. Choose whether submission documents may go to an approved AI service.
Run it onYour cloud Local environment AI model accessApproved model API Private model only
Data and access details Runs in a cloud account you control, with access controls and logs.
Only the submission documents and your guidelines go to the selected external model, under agreed access and retention terms.
Agree retention, access and logs before delivery. Private models keep inference on your hardware; authorised connections to your own systems may still use the network.
How do you want to use it? Use a submission summary page, your current AI agent or your rating workbook. You can choose more than one.
A submission summary page Open each prepared submission with the extracted fields, loss summary, guideline checks, missing items and clearance result, each linked to the page it came from. Included My existing AI agent Let your current agent ask for a submission's status, its missing items or its summary, and read the results. Setup +$72 – 240 My rating workbook Fill the input tab of the rating spreadsheet you already use with the normalised fields, for an underwriter to check before rating. Setup +$144 – 360