Insurance operations
Our team built a certificate-of-insurance workflow for a dispatch network: carrier intake, Sensible OCR, requirements checks and follow-up through resubmission.
Updated 2026-10-25
The challenge
The network needed to know whether each carrier's certificate met its insurance requirements. Reading a PDF was only part of the job. Someone also had to check the extracted details, explain any deficiencies and keep track of the next submission.
Our approach
We connected COI intake to Sensible for document extraction, then passed the parsed fields to our own rules engine. The engine evaluated those fields against the network's requirements. Its result determined the carrier's next step: confirmation, or a list of issues to fix before resubmitting.
Inside the experience
A visual guide to the published product scope.
A COI enters the intake flow.
Document fields become structured data.
Requirements checks identify compliance or deficiencies.
Confirmation closes the loop; deficiencies lead to correction and resubmission.
Conceptual illustration · Based on public capabilities, not a private architecture diagram.
An anonymized account of our team's earlier delivery for a dispatch network. The companies submitting certificates handled dispatch operations for that network. Client and partner identities, actual requirements and customer documents remain private.
Our contribution covered the carrier submission flow, Sensible OCR integration, custom requirements evaluation and the carrier communication loop. We connected these pieces so the result of a document check led to a specific next action.
The delivered workflow connected document intake to a requirements decision and the next carrier response. A carrier with deficiencies had something specific to act on and a route back into the same process. The submission, result and follow-up belonged to one workflow.
Separating extraction from the requirements engine gave each part a clear responsibility. OCR supplied field values; business rules evaluated them. A failed check also had to be useful to the carrier, because a status alone wouldn't explain what to fix. The public demo below illustrates that distinction with fictional documents and a small teaching rule set.
Try the workflow
We rebuilt a small version for the Rubix site with fictional sample certificates. Submit an ACORD COI, inspect the result, then switch to the admin workspace.
The demo adds a deliberately ambiguous extracted amount, editable teaching requirements and visible decision history. These illustrate how a similar tool can work today; they aren't a claim about the original client's exact interface or controls.
Rubix can help work through your intake, extraction choices, requirements and carrier follow-up. The build guide covers the implementation questions; a project conversation starts with the process your team needs.
All screenshots show fictional data. The demo uses simulated extraction, sends no emails and makes no insurance acceptance decision. The original client system and source code are private.
Related case study
We built an in-house OCR workflow that extracts driver license data, reduces manual entry, and gives operations teams a review step.
Rubix Labs
A few finishing touches.
