Two-week onboarding became three days with AI Copilot
An anonymous Hong Kong insurance brokerage used a grounded Dyna.Ai Copilot to retrieve approved policy knowledge and help advisers prepare customer-ready answers. Licensed advisers remained responsible for recommendations, suitability, and customer decisions.
The business challenge was the distance between a customer question and the approved information needed to answer it. Insurance advisers work across products, policy conditions, exclusions, and customer circumstances. A question that sounds simple can require several pieces of policy knowledge, and the correct answer may depend on the customer’s situation.
Manual search creates two risks. It slows the conversation while the adviser looks through multiple sources, and it can create inconsistency when different advisers use different starting points. The brokerage did not need a generic answer tool. It needed a way to bring approved knowledge into the workflow without weakening the adviser’s role.
The challenge
Advisers needed trusted answers at customer speed
Dyna.Ai was deployed as a grounded AI Copilot connected to the brokerage’s approved knowledge sources. The Copilot could retrieve relevant policy information, prepare a response or starting point, and show the source context for adviser review.
The solution addressed a specific operational need: shorten the path from question to trusted information. It did not replace the licensed adviser and it did not turn a generated answer into a recommendation automatically. The adviser remained the person who interpreted the customer’s circumstances, applied professional judgment, and decided how to communicate the next step.
This division of labour is central to the case. The AI handled retrieval and preparation. The adviser handled interpretation, suitability, recommendation, and customer communication. That made the workflow useful for a regulated environment without presenting the assistant as an autonomous insurance adviser.
The deployment
Grounded policy knowledge inside the adviser workflow
The Copilot was placed inside the broker service workflow rather than offered as a separate destination that advisers had to search independently. The available evidence supports approved policy knowledge, product information, company information, and adviser-led support. The exact source systems, access permissions, update ownership, audit process, and deployment phase still need confirmation.
The workflow can be understood as a controlled loop. A customer or adviser raises a question. The Copilot retrieves relevant knowledge. The adviser checks the source and adapts the answer to the customer context. If the request requires judgment or a decision, the adviser remains accountable. If the information is outside the approved source set, the workflow should expose that boundary rather than create false confidence.
How the workflow changed
From policy search to a ready-to-review response
Before the Copilot, advisers could spend time searching across product and policy information before they were ready to respond. After deployment, the adviser received a more direct starting point with source context attached.
The reported case results were a reduction in onboarding time from two weeks to three days, a reduction in response time from two hours to five minutes, and 50% night-time query resolution. The business change was a faster path from customer question to adviser-ready response, without removing the licensed adviser from the decision.
The results
Faster onboarding, answers, and after-hours resolution
The brokerage gave advisers a faster, more reliable path to trusted information while preserving licensed human responsibility. The Copilot supported the service workflow; it did not provide regulated advice independently.
