A three-part series by OpenDialog and Durell Software
From Queries to Conversations – Turning Client Interactions into Structured, Actionable Data
In Part 1 of this series, we explored how AI-driven document ingestion can turn PDFs and attachments into structured data that flows cleanly into Durell.
But for most brokers and MGAs, documents are only one side of the story.
A large proportion of operational work still comes through everyday client interactions – phone calls and emails:
- querying existing cover and other covers available
- requesting policy documents or clarifications of content
- making mid-term cover adjustments
- confirming renewal instructions
- notification of loss
These interactions are essential — but they are also one of the most significant sources of administrative effort. Moreover, information arrives in different formats, at various levels of detail, and with varying levels of accuracy. The data must then be interpreted, clarified and eventually rekeyed into internal systems.
In 2026, Durell and OpenDialog plan to integrate OpenDialog’s conversational AI agents into customer-facing workflows to help structure and standardise these interactions.
The problem: unstructured conversations create operational friction
Client communication generates several recurring challenges:
1. Time-consuming data collection
Collecting the right information for an MTA, renewal or claim often requires follow-up questions and repeated back-and-forth with the client. As a result, handlers lose time that could be spent on higher-value work.
2. Variability between handlers
Different members of staff may ask questions differently or interpret responses differently. Consequently, this leads to inconsistencies in the recorded data.
3. E&O (Errors & Omissions) exposure
Free-flowing conversations can lead to missing details or explanations phrased inconsistently. Therefore, brokers face higher operational and compliance risk.
4. Data that is difficult to reuse
Any information shared via email or phone must be formatted and manually entered into systems. In addition, this results in a high administrative burden and potential for human error.
AI + Durell: introducing structured conversational workflows
Durell’s first AI agent is designed to support brokers by enabling guided, rule-based conversations that ensure the correct information is consistently captured and structured.
Using Natural Language Processing and broker-defined workflows, the AI agent can:
- guide clients through predefined question sets for MTAs, renewals and claim notification
- validate responses for completeness and clarity
- provide standardised explanations to improve consistency
- produce structured summaries for staff to review
Importantly, as with Part 1, human oversight remains central.
Staff are required to review and amend or confirm the information captured by the AI agent.
Staff are required to review and amend or confirm the information captured by the AI agent. Consequently, professional judgment, regulatory requirements and service quality remain firmly in the hands of humans.
Stronger auditability and governance
One of the key benefits of structured conversational workflows is the ability to create clearer, more consistent records.
In addition, the AI agent will support governance by:
- ensuring the same questions are asked every time
- capturing client responses in a structured, auditable format
- reducing the risk of missing information
- supporting good internal controls and market-compliant processes
Overall, these improvements help reduce operational risk while enhancing the quality of data held within Durell.
Expected impact for brokers and MGAs
By reducing variability and providing clearer, more complete information, the AI agent has the potential to deliver meaningful operational improvements:
- Lower administrative workload — fewer ad-hoc emails and clarifications
- Improved data quality — cleaner, structured information ready for review
- More consistent client experience — the same standard every time
- Stronger governance — records that support compliance and audit requirements
Therefore, these enhancements support both service excellence and operational efficiency.
The takeaway
Part 1 demonstrated how AI can transform documents into structured, reliable data.
Similarly, Part 2 shows that the same can be achieved for everyday client interactions.
By combining Durell’s industry expertise with OpenDialog’s conversational AI agent, brokers and MGAs will be able to capture information more consistently, reduce their administrative burden, and enhance the quality of the data flowing into their systems.
Finally, in Part 3, we will explore how Durell plans to build on this foundation — using structured data across documents and conversations to improve productivity and introduce deeper automation with OpenDialog’s AI agents.
Durell provides the Policy Administration System and Underwriting Platform that brokers and MGAs already trust.
OpenDialog adds AI that reads, understands, and automates data movement through those systems.
Together, they deliver a more innovative way to handle insurance — from document ingestion to automated customer service — compliant from the ground up.
Frequently Asked Questions
AI – Conversational Workflow
AI uses guided question sets and validation rules to ensure the right information is collected consistently, reducing back-and-forth and improving data quality.
OpenDialog provides the conversational AI engine that guides clients, validates responses, and produces structured summaries for staff to review inside Durell.
Yes. As with document ingestion, human oversight remains essential. Staff confirm or amend captured information to ensure accuracy and maintain compliance.
Client questions, MTAs, renewal confirmations, claim notifications and cover clarifications can all be structured and standardised using conversational workflows.
Durell and OpenDialog plan to introduce the document ingestion in Q1 2026 and the conversational AI workflows in Q2 2026, supporting brokers and MGAs with more consistent and efficient processes.