Show HN: Content quality analysis that finds what SEO tools miss

Posted by timsul on Monday October 27, 2025 @09:38AM
from the show-hn dept.

Hi HN! I'm Tim! We rejected 73% of our AI-generated content last month. Not because of grammar, but because it was substantively empty. AI made content production 10x faster, but quality checking didn't scale. I built FluffFilter to catch this before publishing.

Content teams went from producing 25 posts/week to 250+, but were publishing a lot of generic fluff—phrases like "in today's digital landscape," unsubstantiated claims, surface-level analysis.

Existing tools (Grammarly, Clearscope, etc.) focus on mechanics and SEO keywords, not substantive quality.

What FluffFilter Does:

  * Analyzes content against 20+ specialized evaluators
  * Automatically detects content type (blog post, case study, email, etc.)
  * Returns 3-5 surgical fixes with exact locations and suggested replacements
  * Batch processing for reviewing 50+ documents at once

Technical Stack:

  * Rails 8.0 + SQLite: Decided against starting with Postgres + Redis to launch with zero-dependency stack. Deploys in 30 seconds, is cheap to host.
  * Solid Queue replaced Redis: One less service to manage, job processing works identically.
  * Claude Sonnet 4.5: ChatGPT did pretty well too, but found Claude's analyses felt better.
  * Turbo Streams for real-time updates

Example Analysis:

I analyzed an AI-generated "marketing trends" blog post. It would score well on SEO tools (good keywords, readable), but FluffFilter found: * Zero concrete examples or verifiable data * Vague audience (unclear who it's for) * Ends with "Would you like that?" like a personal email

See the full analysis: https://flufffilter.com/examples/blog-post

More examples: https://flufffilter.com/examples

Try it yourself: 7-day trial with 15 analyses at https://flufffilter.com

Built this nights and weekends. Would love HN's thoughts on:

  1. The technical approach
  2. Whether the AI feedback quality is genuinely useful vs. generic
  3. What other content types would be valuable to evaluate

Launching on Product Hunt today as well, but genuinely curious what HN thinks.

Re:Show HN: Content quality analysis that finds what SEO too... by sanskarix on Friday October 31, 2025 @03:10AM (#45769158)

genThis is super interesting! The technical stack is impressive, Rails 8 + SQLite with zero dependencies is exactly the kind of lean approach that makes sense for launching quickly. The Solid Queue choice over Redis is particularly smart for keeping costs down.

I'm curious about your Product Hunt launch strategy. Did you build up a network of supporters beforehand, or are you launching cold? I'm launching Cal ID (a scheduling/identity tool) on Product Hunt on November 10th and would really appreciate any tips on what worked for you. The AI content quality checking angle seems like it would resonate well with the PH audience.

Also wondering, have you considered analyzing social media posts? I could see this being valuable for LinkedIn content where generic fluff is everywhere.

[ Parent | on HN ]