Submissions in. Structured data out.
Broker emails, ACORD forms and loss runs become reviewed, structured records in your policy admin system in about four minutes instead of forty-five — so your underwriters price risk instead of retyping it. Built for MGAs, program administrators and wholesale brokers.
30 minutes · run on one of your own submissions · nothing to install
In production today
The intake problem hasn't been solved. It's been tolerated.
Most MGAs run submission intake on email threads, shared inboxes and copy-paste. It is a structural bottleneck, and it gets worse exactly when business is good.
Your most expensive people are typing
45 to 90 minutes per submission, spent by licensed underwriters extracting ACORD fields, loss-run rows and vehicle schedules by hand. That is not underwriting capacity, and it is the most costly data entry on the org chart.
The broker already went somewhere else
Email intake has no queue, no status and no owner. While a submission waits its turn to be keyed, the retailer is getting an indication from a market that answered first. The loss never appears in a report, because nothing was ever tracked.
Every re-key is an E&O exposure
A transposed VIN, a missed loss-run row, a misread ACORD checkbox. Each one risks a mis-rate or a coverage dispute, and none of it is auditable afterwards — there is no trail back from the record to the page it came from.
What actually comes out the other side
A typed record, every field scored, and anything uncertain routed to a person instead of guessed.
- Named insured
- Ridgeline Freight LLCHigh
- DOT number
- 3184722High
- Effective date
- 2026-09-01High
- Power units
- 42High
- Radius of operation
- 500+ milesHigh
- Prior carrier
- illegible on p.7Review
One field could not be read confidently, so it is flagged rather than guessed. A person resolves it before anything posts to your policy admin system.
Illustrative example built from representative field values. Not a customer record.
From inbox to policy admin in four steps
Every submission follows the same auditable path — no black-box magic, no data sitting in a silo.
Ingest
Broker emails and their attachments arrive at the address you already publish. RM Assistant picks them up automatically — no forwarding rules, no upload portal, nothing for your brokers to adopt.
Classify
Every page is identified: ACORD 125, 127 and 140, loss run, schedule of vehicles, MVR, supplemental. Confidence is scored per document, so the pipeline knows what it is unsure about.
Extract
Insureds, drivers, vehicles and claims are pulled from every document and typed against your field schema. VINs are decoded. Every field carries a confidence score and a link to its source page.
Review, then push
A person approves the record — correcting anything flagged — and only then is it written to your policy administration system. Nothing posts unreviewed.
What it changes for operations
Measured on live commercial submissions, including commercial trucking — the messiest stack in the book.
Figures are from production commercial-lines programs, measured end to end from inbound email to a posted record. Your document mix will move these numbers, which is why we benchmark on your own historical submissions before you commit.
The questions every ops leader asks
Usually in the first ten minutes, and usually in this order.
What happens when it is not sure?
Low-confidence fields are flagged, never guessed. They surface at the top of the review queue with the source page alongside them. The failure mode is a human look, not a silent wrong value in your policy admin system.
Who owns the data?
You do. Submissions and extracted records are isolated per customer and pushed into your own systems. Nothing is pooled across customers, and submission data is never used to train shared AI models.
Can we audit it two years later?
Every field keeps a pointer to the document and page it came from, together with the confidence score and any human correction. When a rating question or an E&O query arrives long after the fact, the trail is intact.
Do our brokers have to change anything?
No. They keep emailing submissions exactly as they do now. Any system that requires brokers to adopt a portal is a system they can decline by sending the risk to a market that does not ask.
Lands clean in the system you already run
Data flows into your policy administration system — it doesn't sit in ours. Status is shown per system, because an integration that only exists on a website is one you find out about during the proof of concept.
Don't see yours? We build custom integrations.
Submission intake automation, answered
What is RM Assistant?
RM Assistant is submission intake automation for commercial insurance. It receives broker submissions by email, identifies and reads each attached document — ACORD forms, loss runs, vehicle schedules, MVRs — extracts the underwriting data with a confidence score on every field, and writes reviewed, structured records into the customer's policy administration system. It is built for MGAs, program administrators and wholesale brokers.
How is it different from OCR or a general document AI tool?
Generic OCR returns text. RM Assistant returns typed, validated insurance data: it knows what an ACORD 127 is, that a loss run contains claim rows that must be itemised and merged across carriers, that a VIN decodes to a specific vehicle which should reconcile against the count on the application, and that the finished record has to satisfy your policy administration system's schema. That insurance-specific layer between raw text and a postable record is the product.
Does a human still review the data?
Yes, always. Extracted data lands in a review queue with field-level confidence and links to the source document. A person approves or corrects before anything is written downstream. The aim is to remove the typing, not the underwriting judgement.
How long does it take to get running?
A typical program is processing live submissions within a few weeks. The work is mapping your field schema and connecting your system of record, not a multi-quarter transformation. Onboarding starts by running your real historical submissions through the pipeline so you can measure accuracy on your own book first.
Which policy administration systems does it support?
Dayd ALIS is live in production. Vertafore IMS is in development with a design partner. Guidewire and other systems are built against your specific instance via a custom API integration during onboarding. The integrations page states the status of each explicitly.
What does it cost?
Pricing is per submission processed rather than per seat, so cost tracks volume instead of headcount, and reviewer accounts are free. The rate depends on monthly volume and document mix — a monoline submission and a trucking submission with a 400-unit schedule are genuinely different units of work.
What kinds of insurance does it handle?
US commercial lines, with the deepest coverage in commercial trucking and commercial auto — the book with the largest vehicle schedules and heaviest multi-carrier loss runs. The same pipeline handles broader commercial lines using the relevant ACORD forms and supplementals.
See it run on one of your own submissions
30 minutes. We take a real submission from your book — email, ACORD, loss run — and run it through the full pipeline while you watch. No slides, no pitch deck.
- 1You send one representative submission
- 2We run it live and show every field
- 3You get the accuracy numbers in writing
No sales pressure. We confirm within one business day.