New essay → How I'm building Ouralink in public
Featured AI Automation Project

Insight Engine

Turn raw customer feedback into product decisions.

Insight Engine Voice of Customer dashboard showing feedback records, theme clustering by size and sentiment, and a priority map

Why

Customer feedback is everywhere — and nowhere useful.

Product teams get signal from reviews, tickets, surveys, and interviews — then someone has to read it all. Insight Engine replaces manual synthesis with an AI pipeline so decisions are objective, evidence-backed, and fast.

What

One engine — from raw feedback to product intelligence.

A full-stack AI product that ingests unstructured feedback, clusters themes with semantic embeddings, scores sentiment and urgency, and ships executive-ready Voice of Customer briefs — all in an interactive dashboard.

Latest Voice of Customer brief generated by Insight Engine with themed sections, sentiment, urgency, and evidence quotes

How

An end-to-end AI pipeline.

Every step is composable, observable, and production-ready — from ingest to executive-ready report.

01

Ingest

CSV, JSON, or REST API feedback streams.

02

FastAPI Backend

Typed pipelines with SQLite / PostgreSQL persistence.

03

Embeddings

MiniLM sentence embeddings with a TF-IDF fallback.

04

Semantic Clustering

HDBSCAN discovers themes without predefined categories.

05

Sentiment & Urgency

Score every cluster for tone and priority.

06

Claude Summaries

AI-written executive briefs, with template fallback.

07

Streamlit Dashboard

Interactive KPIs, themes, quotes, and drill-downs.

08

VoC Report

Board-ready Voice of Customer, generated in minutes.