Live · repreport.vercel.app · on hold

Social listening into sentiment reports

Brand mentions scatter across Twitter, Reddit, LinkedIn, Instagram, Facebook, with no single lens. Define a study, pull mentions per keyword, classify sentiment and feature requests, render a report.

reel · promo
reel · demo
date
Oct · 2025
the problem →

Brand mentions scatter across Twitter, Reddit, LinkedIn, Instagram, Facebook, with no single lens.

what I built →

Define a study, pull mentions per keyword, classify sentiment and feature requests, render a report.

The story

Every brand I’ve worked with has the same blind spot. People are already talking about them on Reddit, LinkedIn threads, Instagram comments, Twitter replies. Nobody on the team has a single place to see it. Marketing watches one corner, support watches another, product sees none of it.

I didn’t want a live dashboard that nobody opens. I wanted a snapshot you can share: the kind of thing you drop into a Monday standup or a board update and actually read top to bottom.

So I built RepReport. You describe the brand once, it goes and listens, and hands back a report.

What it does

You create a study in a 3-step wizard: company name, platforms to watch, and what you care about (praise, complaints, feature requests). RepReport pulls real mentions and classifies each one with a structured-output LLM call so the sentiment bucket is deterministic. Then a second pass reads the top mentions and their comment threads to write the narrative summary, bilingual EN/ES, with citation links baked into the prose back to the original threads.

The output is a single report page per study:

keep exploring →
Other chapters in this notebook
Back to index