Building a generative AI backend for analyzing customer feedback

Posthog is a product analytics platform that, among its many features, allows for embedding freeform surveys on a homepage.

Freeform surveys (where users can type anything into a text box) capture a ton of data, but are hard to apply metrics to. This is a perfect use case for generative AI.

In this project, I built a system to capture the freeform survey inputs and analyze them using sentiment analysis and using natural language queries. For example, one could ask: 'What do people think of Product B - Feature A' or something more generally like 'Are our customers happy?' and get actionable data back from the LLM.

More details below on Github.