A three-part series by OpenDialog and Durell Software
Part 3: From Structure to Productivity – Building a Governed Foundation for Automation
In Part 1, we explored how AI-driven document ingestion can convert PDFs and attachments into structured, usable data within Durell.
In Part 2, we examined how the same principles can be applied to everyday client conversations — introducing structure, consistency and clarity to emails and phone calls through guided conversational workflows.
Together, these two capabilities address a fundamental challenge in insurance operations: much of the industry’s most important information still arrives unstructured.
Part 3 looks ahead. Not to launch new products or make bold promises — but to explain how structured data, captured consistently across documents and conversations, creates a practical foundation for improved productivity and carefully governed automation.
From isolated efficiency gains to a connected foundation
Document ingestion and conversational workflows each deliver value in their own right. However, their real impact emerges when they are combined.
When data from proposal forms, endorsements, MTAs, renewals and claims notifications is captured in a consistent, structured format, it becomes significantly easier to review, validate and reuse. As a result, staff spend less time rekeying or reconciling information and more time applying professional judgment.
This is the foundation on which Durell plans to build its approach to AI-assisted productivity.
Introducing the Durell Assist family
Rather than positioning AI as a single system or autonomous decision-maker, Durell plans to develop a family of focused, workflow-driven capabilities, referred to collectively as Durell Assist.
Durell Assist is not intended to replace brokers or underwriting teams. Instead, it is designed to support them — helping structure information, reduce duplication, and ensure that data flows through systems in a more consistent and controlled way.
Each capability within the Durell Assist family is planned to be:
- Purpose-specific
- Subject to human review and confirmation
- Aligned with regulatory and governance requirements
This approach reflects Durell’s view that AI should be applied incrementally, transparently, and only where it delivers genuine operational value.
Tona: the conversational layer
As explored in Part 2, structured conversations play a critical role in capturing accurate information at source.
Within the Durell Assist family, Tona is the planned conversational workflow capability.
Tona is designed to guide clients through common insurance processes — such as Policy queries, document requests, MTAs, renewals and notifications of loss — ensuring that information is captured clearly and consistently.
Tona — born by the River Tone, built by Durell.
Importantly, Tona is not positioned as a virtual broker or autonomous agent, it is a productivity tool built to support brokers.
Productivity through consistency, not autonomy
With structured data available from both documents and conversations, new productivity opportunities begin to emerge.
These include:
- Reduced rekeying across systems
- Fewer follow-up emails and clarification calls
- Clearer handovers between teams
- More consistent application of business rules
Over time, this consistency allows carefully controlled automation to be introduced — always within defined workflows and always subject to human oversight.
Crucially, automation is treated as an outcome of good data, not a goal in itself.
Governance remains central
Across all planned capabilities, governance is a core design principle.
Human review remains mandatory.
Rules are defined and outputs are explainable and auditable.
By ensuring that AI-assisted workflows operate within these boundaries, Durell aims to support operational efficiency without compromising regulatory compliance, professional judgment or service quality.
The takeaway
Part 1 showed how AI can structure data from documents.
Part 2 demonstrated how the same approach can be applied to client conversations.
Part 3 shows how these capabilities together create a foundation for improved productivity and incremental automation.
By combining Durell’s deep insurance expertise with OpenDialog’s conversational AI capability, brokers and MGAs will be able to work with cleaner data, clearer workflows and reduced administrative burden — without relying on opaque or over-promised AI systems.
This is not about hype.
It is about building a practical, governed path forward.
Durell provides the Policy Administration System and Underwriting Platform brokers and MGAs already trust.
OpenDialog provides the conversational AI capability that helps structure and standardise interactions.
Together, they are shaping a careful, compliant approach to AI-assisted insurance operations.
Frequently Asked Questions
AI – Conversational Workflow
Durell Assist is a planned family of AI-assisted capabilities designed to help brokers and MGAs structure information, reduce administrative effort, and support productivity within defined workflows.
Tona is the planned conversational workflow within the Durell Assist family. It is designed to guide clients through common insurance processes.
No. Tona is not an autonomous agent or virtual broker. It is intended as a productivity tool that supports consistency while keeping decisions and responsibility with human staff.
Capturing information consistently from both documents and conversations reduces rekeying, follow-ups and interpretation errors, allowing staff to focus more on decision-making.
All planned AI-assisted capabilities are designed to operate within defined rules, require human review, and produce auditable outputs that support regulatory and compliance requirements.