CleviCS AI Insurance Platform

Context

CleviCS needed a private insurance consultation platform that could support both customer self-service and advisor-assisted workflows. The product had to help customers describe what they needed, guide them through a structured consultation, and give the operations team a practical way to continue the conversation when a human follow-up was needed.

Because this was company work, this case study intentionally stays at a product and workflow level. Source code, internal configuration, provider details, customer data, and implementation specifics are not public.

Problem

Insurance consultation is hard to turn into a clean digital flow. Customers need guidance without feeling boxed into a form, while advisors need enough context to understand the conversation quickly and follow up responsibly.

The product also needed to support voice follow-up from the admin side. That meant the experience could not stop at chat; it had to include call initiation, active-call status, history, summaries, and review workflows.

Approach

I focused on the product surfaces that make the workflow understandable:

• A customer consultation room where an AI assistant can collect needs, ask clarifying questions, and prepare a recommendation path.
• An admin console where managers can review customers, continue conversations, and start outbound call workflows.
• A call-review area where completed conversations can be summarized and checked before advisor follow-up.

The goal was to make the AI feel like part of an operating workflow, not a separate chatbot pasted onto the product.

What Shipped

• Customer-facing AI consultation flow for insurance needs discovery.
• Admin-side workflow for monitoring consultation status and preparing advisor follow-up.
• SIP-oriented outbound-call console for manager-assisted phone consultation.
• Call history and review surfaces for summaries, transcripts, and quality checks.
• Private operational flows designed to avoid exposing sensitive customer data publicly.

Results

• The product connected AI consultation, human advisor handoff, and voice follow-up into one workflow.
• Managers gained a clearer operational surface for deciding when and how to follow up with a customer.
• Completed calls became easier to review because summaries and quality states were visible in the admin experience.

Technical Notes

The implementation involved a modern web frontend, backend APIs, realtime events, AI-assisted conversation flows, database-backed records, and SIP-based call handling. I am intentionally omitting deeper architecture, provider configuration, prompt details, and internal service design because the project is private.