Why this matters now
Insurance distribution through banking channels places unique demands on front-line advisors. At any moment, a bank advisor may be preparing for a customer meeting, handling an inbound phone inquiry, or guiding someone through a product decision in branch. Each scenario requires fast, accurate, and contextually relevant knowledge.
As the insurer's portfolio with the bank grows across personal, business, and specialty insurance lines, static intranets and PDF guides can no longer keep distributed advisor teams consistently informed. Knowledge becomes outdated, guidance varies by branch, and customers experience longer wait times and less confident service.
Advisors need real-time, segment-aware product guidance, but maintaining a living knowledge base across a distributed branch network is operationally complex without the right AI infrastructure.
The company's goals were to:
- Reduce advisor preparation time and increase consistency across all interaction types: preparation, phone call, and in-branch visit.
- Maintain knowledge currency across the branch network through a single, centrally managed AI knowledge layer.
- Measure real usage and quality through custom KPI dashboards and feedback loops tied to Genesys session data.
- Scale from a controlled pilot to the full Banco Sabadell advisor population without disruption.
The Challenge
Supporting bancassurance advisors at scale meant solving for four interconnected needs simultaneously:
- Quick, high-quality sales assistance to position the right products by interaction type and customer segment.
- A dynamic Knowledge Base with metadata-tagged files and URLs, always up to date without manual intervention.
- A fully conversational agent in the local language, capable of handling complex insurance scenarios end-to-end.
- Deployment inside Microsoft Teams via Genesys integration, fitting seamlessly into advisor workflows.
The Solution
Druid AI and the insurer co-built a purpose-built AI Agent for the bancassurance advisor experience, an assistant that understands the full complexity of insurance sales in a banking context, deployed where advisors already work.
- Intelligent Prompt Engine
A structured prompt engine guides advisors through the right product positioning based on three dimensions: the specific insurance product, the target customer segment, and the interaction type. Responses are delivered in consistent, ready-to-use formats. - AI Agent in Microsoft Teams via Genesys
The agent is deployed natively in MS Teams and integrated with Genesys, matching exactly how the bank's advisors work. - Deployment on intranet and WhatsApp
Employees reach the agent through the channel they're already using - the corporate intranet during work hours, WhatsApp from their phone outside the office or on the road. - Dynamic Knowledge Base (GPT-Powered)
A GPT-powered knowledge base ingests structured files and live URLs, tagged with metadata for product line, customer segment, and interaction type. Advisors always access the most current version - no manual updates, no version confusion. - Custom KPI Dashboards & Feedback Loops
Purpose-built dashboards track automation integration rates, a Clarification KPI for measuring advice quality, and custom time-to-answer (TTA) drivers, all correlated for full traceability.
Learnings
Deploy where advisors already work
Embedding the agent in MS Teams via Genesys meant zero workflow change - advisors adopted it naturally from day one.
A dynamic knowledge base is a living asset
With metadata-tagged files and URLs updated centrally, the agent stays current without manual intervention, and knowledge consistency across over 1.000 branches becomes automatic.
Structure beats flexibility in complex sales contexts
Organizing responses by product, customer segment, and interaction type gave advisors immediately actionable guidance, and ambiguity was engineered out from the start.
Advisor-led UAT is non-negotiable
130+ feedback items were caught and resolved before go-live; that rigor was the difference between a confident rollout and costly post-production fixes.
