A three-part series by OpenDialog and Durell Software
From Paper to Productivity – Turning Insurance Documents into Structured, Actionable Data
The insurance industry runs on documents — Proposal Forms, Statements of Fact, Schedules, TOBAs and bucketloads of correspondence.
Every transaction generates data, and much of it still arrives as PDFs and other documents attached to emails.
For brokers and MGAs using Durell, document management and storage are already covered. The challenge is how quickly that information can be converted into usable data — without endless rekeying or compliance risk.
The problem: manual ingestion slows everything down
Reading, typing, and cross-checking data across systems is still one of the biggest time drains in a broker’s day.
Manual ingestion creates three core problems:
- Time drain: Staff spend hours copying data from documents into systems.
- Error exposure: Small rekeying mistakes can cause compliance issues or E&O (Errors & Omissions) exposure — where an administrative error results in client detriment.
- Lost visibility: When key details stay trapped inside attachments, they can’t feed reporting, renewals, or analytics.
AI + Durell: automating data ingestion with human oversight
Durell are integrating OpenDialog’s AI engine into their e-trade marketplace and updated Policy Administration System Durell 2.0 to bring automation and intelligence to this process in 2026.
Using Optical Data Extraction to read scanned documents and Natural Language Processing (NLP) to understand context, the system can:
- Recognise document type (Proposal Form, Statement of Fact, Schedule).
- Extract risk information such as client name, address, cover dates and sums insured.
- Validate that information using OpenDialog’s decisioning layer before matching it against Durell’s Polaris-based data capture screens.
Users are presented with an indication of confidence in the risk information extracted and are required to review and amend or confirm any data automatically captured.
Once confirmed, the data is written into the Durell fields with full audit trail and can be used to generate quotes, reporting and other analytics.
This process keeps humans in the loop, maintaining professional judgment and accountability while dramatically reducing admin effort and likelihood of error.
Compliance and audit strengthened
Every automated extraction and confirmation is timestamped within Durell’s audit trail.
That means:
- Transparent data lineage from document to record.
- Compliance with good internal governance practices.
Improved overall data hygiene and accountability.
Real-world impact
Durell are working with an existing broker to rebroke their book using OpenDialog’s AI document ingestion with human review:
- Processing time has dropped by 80%.
- Accuracy is over 99%.
- Administrative staff have been redeployed to the New Business team.
The takeaway
The partnership between Durell’s industry experience and OpenDialog’s AI expertise is already delivering results. Using AI and built-in human confirmation, firms can safely automate document handling — turning static paperwork into structured, reliable, and compliant data in minutes.
- Durell 2.0 overview – https://www.durell.co.uk/news/introducing-durell/
- Software / Policy Administration page – https://www.durell.co.uk/system/software/
- Contact or demo page (CTA) – https://www.durell.co.uk/contact/
Frequently Asked Questions
AI – Document Ingestion
AI insurance document ingestion uses technologies like Optical Data Extraction and NLP to read, classify and extract key information from documents automatically. In Durell, this process is enhanced by OpenDialog’s decisioning layer, ensuring accuracy with human confirmation where needed.
OpenDialog provides the AI engine and decision framework that routes extracted fields for human review when confidence is low. This ensures data quality remains high while reducing manual rekeying and audit risk.
Proposal Forms, Statements of Fact, Schedules, TOBAs and supporting documents such as emails and PDF attachments can all be processed automatically, turning static files into structured data.
Yes. Each extraction includes a confidence score, and users must confirm or amend fields before the data is committed. This keeps humans in control of the final outcome while providing a full audit trail that supports governance and regulatory oversight.
No. It removes repetitive typing and cross-checking so staff can focus on higher-value tasks such as renewals, new business and client service. In one example, admin staff were redeployed to revenue-generating activities after automation.