Scribbie

Inspiration:

Customer care and sales reps often struggle to multitask during calls — listening, understanding, taking notes, and planning follow-ups. We wanted to give them an AI-powered assistant that handles the heavy lifting so they can focus on the human connection.

What it does:

Scribbe transcribes customer calls, summarizes the conversation, identifies key concerns and requests, and auto-generates a structured PDF report — complete with a to-do list and sentiment-based CSAT score — all in real-time.

How we built it:

We used React for the frontend, Node.js/Express on the backend, and AssemblyAI for accurate transcription. Custom AI prompts drive the NLP extraction pipeline, and sentiment analysis enhances insight. Everything is rendered dynamically and downloadable as a clean report.

Challenges we ran into:

Polling async transcriptions, synchronizing audio with utterances, and ensuring structured JSON output from AI in real time were all tricky — but we tackled each one with smart design choices.

Accomplishments that we're proud of:

We built a working AI support assistant in under 24 hours — end-to-end — from file upload to PDF report. It’s clean, fast, and ready to scale.

What we learned:

How to design a full-stack pipeline with external APIs, prompt engineering for NLP tasks, and creating tools that empower people, not replace them.

What’s next for Scribbe:

Live call integration, CRM embedding, real-time alerts for customer churn signals, and team-level analytics. Scribbe is just getting started — because support deserves support.