Every agency knows the ritual. A policy comes back from the carrier, and someone on the servicing team sits down with the issued document on one side and the proposal on the other, comparing them line by line — coverage, limits, deductibles, endorsements — hunting for the place where what was issued is not what was sold. It is slow, it is repetitive, and it is one of the most consequential hours in the building, because the discrepancy nobody catches in that chair tends to surface later, after a loss, in a dispute about who owes what.
This week that chair got an occupant of a different kind. Vertafore has made its Velocity Policy Check Agent generally available in AMS360, alongside a Policy Intake Agent that reads policy documents into the agency management system. The Check Agent compares the carrier-issued policy against the agency’s records and source documents, highlights discrepancies, and suggests corrections. A human reviews and approves before anything changes; the system creates suspenses where follow-up is required and logs the review.
What the numbers claim, and what they are
Vertafore reports that in early trials the Check Agent reduced policy-checking time by up to 90 percent per line of business, and the Intake Agent cut the work of building and reviewing a standard commercial property and general liability package by up to 87 percent. Those are the vendor’s figures, from the vendor’s trials, on standard packages — the clean end of the book. Agencies running messy, manuscripted, or heavily endorsed policies should expect less, and should notice that the honest word in both claims is “up to.” But the direction is not really in dispute: the comparison step is machine-shaped work, and machines now do it inside the system agencies already run.
Why this is a liability story, not a software story
The policy check is not clerical hygiene. It is the control that stands between an agency and its most classic errors and omissions claim: the client who was told they had coverage the issued policy does not contain. Errors and omissions insurance exists for the day that gap is discovered — and as any broker who has lived through one can tell you, it is usually discovered at the worst possible moment, with a claim already on the table and the claims-made calendar already running.
An agent that checks every issued policy, instead of a fatigued human sampling a stack at the end of a long day, changes the odds on that claim in the agency’s favor. It changes something else, too: the record. Every review is logged. In a dispute, the agency that can produce a documented check of the exact policy at issue is in a fundamentally different position from the agency testifying that checking was standard practice. The audit trail is not a byproduct of the tool. Defensively, it may be the most valuable thing the tool produces.
The new weak point has a name: the approval click
Both agents are built so a person approves the machine’s work before it lands. That design is doing two jobs at once — it keeps a professional in the loop, and it moves the liability to a specific moment: the approval. Which creates the failure mode agencies should plan for now, while the habit is forming. A reviewer who approves flags at speed, trusting the agent the way drivers trust a navigation app, quietly recreates the rubber stamp the policy check existed to prevent. The tool will keep excellent records of that, too.
So the desk’s read is simple. The checking itself is about to stop being the hard part. The judgment about what the machine found — and the discipline to actually exercise it — becomes the job. Agencies that train for that distinction will get the E&O benefit. Agencies that treat the agent as a reason to stop looking will eventually read their own audit log in a deposition.
Related reading: Errors and omissions insurance · Claims-made vs. occurrence