The Challenge
Our product and ops teams were manually analyzing over 50,000 Zendesk support interactions monthly. This process was slow, and we were missing real-time insights into critical customer pain points.
The Solution
I took the initiative to design and build an end-to-end, serverless AI pipeline. I used AWS Lambda and EventBridge to process the interactions in real-time, integrating with the Gemini API to automatically extract, classify, and analyze the sentiment of each customer issue. The data was then stored in DynamoDB.
⚡ The Impact
Built a real-time dashboard on top of this data that provided immediate visibility and drill-down capabilities. This empowered our product team to move from slow manual analysis to instant, data-driven decisions, which directly informed our product roadmap and feature prioritization.