Loss runs chased,
carrier PDFs read into the file.

1. A submission needs five years of loss history from a carrier the insured left.
2. Someone writes the request email, guesses at the right address, and hopes.
3. Two weeks pass. Nobody is tracking which requests are outstanding.
4. A 30-page PDF arrives, and a CSR reads it to pull claim counts and paid amounts.
5. Dec pages and carrier documents pile up the same way, attached but never read into data.

• ✓Loss-run requests drafted from the policy record and sent to the right carrier desk
• ✓Outstanding requests chased on a schedule, with a live list of what is still open
• ✓Loss runs parsed into claim counts, paid and reserved amounts, and loss ratio
• ✓Dec pages and carrier documents read into structured fields on the account

~7 hours saved · every week

Request drafting

Requests are built from the policy record: insured name, policy numbers, terms, and the years you need, addressed to the loss-run desk that carrier actually uses rather than a general inbox.

Chasing

Outstanding requests get followed up on your schedule instead of when someone remembers. The producer can see what is still open and how long it has been open, without asking.

Loss-run parsing

What comes back becomes data: claims by year, paid and reserved, open versus closed, and the loss ratio your underwriters will compute anyway. The source PDF stays attached so anyone can check the number.

Dec pages and carrier documents

Declarations pages, endorsements, and carrier correspondence are read into structured fields on the account, so the file reflects the policy rather than just storing a copy of it.

Checks before it lands

Documents for the wrong insured, years that do not match the request, and totals that do not reconcile are held for a person. A summary that quietly omits a year is worse than no summary.

Monitoring

Carrier document layouts change. We catch it from monitoring and push a fix, usually before your team sees a parse come back empty.

Can you request loss runs on our behalf?

Yes. Requests go out from your agency, in your name, with your signature block, using the authorisation the insured has already given you. We draft, send, and chase; the relationship stays yours.

How accurate is the parsing on a messy PDF?

Good, and checked. Totals are reconciled against the document rather than trusted, and anything that does not add up is held for a person. The source PDF stays attached to the account so any figure can be traced back in one click.

Is this just AI?

Reading an unstructured document is exactly where an LLM earns its seat, so yes, that part is AI. Writing values onto the account is deterministic and rule-checked, because a hallucinated paid-loss figure is not a small problem. Your runbook lists which is which.

Do we have to change AMS?

No. We sit on top of the AMS you already run (HawkSoft, NowCerts, AMS360, EZLynx, Applied Epic, QQCatalyst) and write to the fields your team already reports on.

Can we try the document side before committing?

Yes. Our free loss-run request tool drafts the letter for you, no signup, and gives you a sense of the plumbing underneath before you book anything.

What does it cost?

It starts free: a 20-minute call and a written automation map of the 3 to 5 workflows worth automating first. Then a 14-day pilot for $2,499 one time: one workflow, fixed scope, live and used by your team by day 14, or you don’t pay. After that, $2,499 a month keeps it running (monitoring and fixes included) and adds one new automation each month; $4,499 a month runs two build lanes. Month to month, cancel anytime, no setup fees. No per-document charges.

Bring the loss-run pile eating your week.

Twenty minutes, no deck. Show us the task, we tell you whether we can automate it and what it would cost. If the answer is no, we’ll tell you who can.