Novus Insight Analyzer

## Inspiration

As product managers and researchers, we conduct multiple user discovery sessions every week. However, manually parsing a 40-minute plain text interview transcript to extract actionable insights is incredibly time-consuming. Critical feedback frequently gets buried in long documents or lost in Slack channels, leading engineering pipelines to build features misaligned with actual customer problems. I built this tool to collapse that gap.

## What it does

Novus Insight Analyzer takes messy, raw user interview transcripts and instantly translates them into a structured semantic report. In under 30 seconds, it automatically highlights:

• Prioritized Pain Points: Grouped neatly by severity with attached verbatim quotes as evidence.
• Key User Testimonials: Categorized automatically by feature taxonomy tags.
• Jobs-To-Be-Done Mappings: Framing raw complaints into actionable product-requirement frameworks.
• Emotional Signal Tracking: Mapping out sentiment changes throughout the dialogue.
• Suggested Follow-Ups: Deep-dive discovery questions for the next call.

## How we built it

The application leverages a high-performance web architecture:

• Frontend UI Framework: Next.js 14 (App Router) styled with a clean utility-first layout via Tailwind CSS and smooth micro-interactions.
• AI Analysis Pipeline: Powered directly via edge routes communicating with the Google Gemini API (gemini-2.5-flash) for fast data-structure generation chains.
• Metrics & Analytics: Integrated directly with the Novus (Pendo) Web SDK tracking script to monitor active user sessions and click metrics safely.

## Challenges we ran into

One of our biggest hurdles was keeping our scope controlled during a 48-hour sprint. We originally intended to integrate drag-and-drop audio file uploaders and persistent database syncing, but realized it would sacrifice the stability of the core tool. We also had to work through strict production build limitations, debugging unescaped JSX literals and script structure requirements to ensure a green deployment build on Vercel.

## Accomplishments that we're proud of

We successfully shipped a beautiful, responsive, and completely functional single-page AI tool from scratch in less than two days. Better yet, the application accurately maps messy conversational tangents into precise product management frameworks (like JTBD) without any hallucination or latency bottlenecks.

## What we learned

This sprint reinforced a timeless product management truth: shipping a polished, highly reliable, and stable MVP is infinitely more valuable than presenting a half-built matrix of overly complex features. Ruthless prioritization was our key to hitting the deadline.

## What's next for Novus Insight Analyzer

The post-hackathon roadmap includes:

1. Adding native .mp3 and .wav audio upload support with automated background whisper transcription.
2. A single-click "Export to Jira/Notion" webhook extension to send synthesized cards straight into engineering backlogs.
3. Multi-transcript comparison maps to surface shared trends across 10+ user interviews simultaneously.

Built With

• google-gemini-api
• lucide-react
• next.js-14
• pendo-(novus-sdk)
• react
• tailwind-css
• typescript