-
-
Browser view
-
Devpost
Inspiration
Most founders think they understand customer feedback because they read support tickets or app reviews. But the most important signals are usually hidden across Reddit threads, GitHub issues, Discord messages, sales call notes, support emails, and scattered app reviews. By the time the pattern becomes obvious, customers may already be churning.
What it does
Hidden Reviews AI turns the Domain Roulette domain hidden.reviews into a real customer-intelligence product. A team pastes a mixed feedback bundle, and the agent:
- extracts hidden customer evidence
- weights shadow channels like Reddit, GitHub, support, and sales notes
- clusters feedback into product-risk themes
- calculates a hidden signal score
- generates a founder-ready action plan
- exports JSON, CSV, Markdown, and dashboard evidence
In the demo, the agent analyzes NimbusLedger and finds a 97/100 hidden signal score, with pricing trust risk as the top issue.
How I built it
I built a deterministic Python agent engine, a static HTML/CSS/JavaScript dashboard, a local API, test coverage, and exportable evidence artifacts. The system runs without paid API keys so judges can reproduce it immediately.
Challenges
The hardest part was making the project feel like a real product instead of a simple summarizer. I focused on preserving evidence quotes, ranking patterns clearly, and making the domain connection obvious: hidden.reviews finds the reviews customers never put in one place.
What I learned
I learned how to turn a random domain into a product thesis, scoring model, dashboard, and SaaS-style workflow that could realistically help founders and product teams.
What's next
Next steps are live integrations for Reddit, GitHub, Discord, app stores, and support tools, plus saved workspaces, scheduled founder digests, and team collaboration.
Log in or sign up for Devpost to join the conversation.