PromptLens — Quality badges for AI-written answers

Annotate content type and evidence density on Zhihu, forums and long-form pages

Problem

Many readers feel that upvotes and quality are two different things—emotional storytelling outruns fact-based answers. A developer already built a browser extension that uses AI to judge content type, evidence density and conflict-of-interest suspicion next to each answer, proving both demand and willingness to bring one's own key.

Solution

A lightweight browser extension that judges each post as you scroll, calling a user-configured LLM to output content type (analysis / personal story / emotional rant / joke / soft ad), evidence density 0–10, and conflict-of-interest suspicion 0–1. Content hashing caches results to avoid duplicate billing, keys stay in local storage, and the page is never pre-fetched.

Users

Readers of Zhihu, Reddit and HN long-form threads, researchers, and content operators

MVP

Scroll-triggered lazy judging; three badge types; content hash cache; user-supplied API key; support Zhihu plus one or two major forums

Revenue

Free basic tier plus paid multi-site support and custom dimensions; team tier with shared API key and usage dashboard

Why now

AI-generated content and emotional upvote farming are both exploding, readers urgently need a 'should I trust this?' signal, and general platforms won't build vertical quality checks for them