Vision Service Plan

About This Project

We integrated Google Cloud conversational AI into a digital vision insurance site so visitors could get instant answers about their plans instead of waiting on hold. The client runs a major portal where families buy and manage eye care benefits online. They needed real-time help without bogging down backend servers — so we brought in Dialogflow CX to answer questions, walk users through plan options on the page, and generate valuable data insights from every interaction. Built over 8 weeks for a USA healthcare client, this project demonstrates how conversational AI in healthcare can streamline customer support through expert healthcare chatbot development.

Challenges

• •Support agents were overwhelmed by the same daily questions about pricing, coverage, and eligibility.
• •Visitors got stuck figuring out which plan details mattered, leading to abandoned carts.
• •Hiring more human agents to handle spikes in basic questions was not financially sustainable.
• •Users rarely typed full sentences — vague queries like "cheap plan?" or "family?" confused existing search tools.

Solutions

• •Mapped conversational logic in Dialogflow CX with specific triggers and routes so the bot understands user intent — forming the backbone of an effective AI customer support chatbot.
• •Built a lightweight embedded chatbot widget from scratch using HTML and JavaScript that sits on the site and talks directly to the Dialogflow backend.
• •Pushed all chat logs into GCP BigQuery for later analysis and meaningful data insights, with Cloud Logging for real-time debugging.
• •Trained the bot on hundreds of poorly typed phrases to improve intent recognition; added fallback flows that suggest helpful links when the AI is unsure.
• •Implemented proactive contextual assistance — if a user sits on the pricing page too long, a modal offers help with plan math.
• •Designed Dialogflow CX to handle messy inputs and typos, turning half-sentences into helpful answers through advanced intent recognition.
• •Tracked page context so the system offers help when users appear stuck instead of forcing them to find a contact button.
• •Built step-by-step conversational guides that ask follow-up questions to walk users through tricky eligibility rules.
• •Planned insurance conversation paths in Dialogflow with a focus on conversational AI for insurance scenarios.
• •Wired up the Google Cloud backend so the client could track chat history and monitor server health as a fully integrated Google Cloud chatbot solution.

Results

• •75% first-touch resolutions — more than three-quarters of visitors get their problem solved by the bot alone, keeping them out of the human support queue.
• •40% boost in time-on-site — people stay longer because they are not bouncing out of frustration when they cannot find a price point.
• •100% chat data visibility — the team went from guessing what users wanted to having complete logs of every question asked on the site.
• •50% drop in agent overhead — filtering out repetitive noise cut daily manual support ticket costs in half.