UntoldAi
UntoldAi

The most welcoming AI product community. Discover, launch, and vote for the best AI tools and LLM wrappers globally.

Built for the Future

Product

  • Browse Tools
  • Market Signals
  • Compare Sites
  • MCP Server
  • AI Agents Hub
  • API Reference

Profession

  • For Lawyers
  • For Students
  • For Teachers
  • For Doctors
  • For Accountants
  • For Real Estate
  • For Recruiters
  • For Marketers
  • For Designers
  • For Developers
  • All AI Hubs

Company

  • About Us
  • Blog Feed
  • Pricing Plans
  • Advertise Placement
  • Affiliates

Makers

  • Submit a Tool
  • Launch Guide
  • For Builders
  • Partnerships

Legal

  • Privacy Policy
  • Terms of Service
  • Platform Rules

© 2026 UntoldAi Inc. All rights reserved.

Made with by makers for makers.

Back to Signals Hub
Data Science & Transparency

Our Measurement Methodology

UntoldAi signals are built on five transparent pillars. No black boxes, no pay-to-rank. Here's exactly how we measure what's winning in the AI tools space.

Signal 01

Community Votes

Every upvote on UntoldAi is cast by a real, registered builder — not a bot, not a purchased click. When a builder upvotes a tool, that signal is logged with a tamper-proof timestamp, deduplication hash, and user fingerprint.

One vote per tool per account — no stacking
Votes are revocable; retractions are instantly subtracted
Authenticated users only — no guest voting
Bot-resistant: rate limiting + session validation on every cast
Signal 02

Bayesian Rating

Raw averages are easy to game. A single 5-star review shouldn't beat a tool with hundreds of 4.8-star reviews. We use a Wilson score combined with a Bayesian prior mean to normalize ratings across tools with different review volumes.

Prior weight: 5 (baseline confidence anchor)
Prior mean: 3.5 (expected baseline quality)
Formula: (prior_weight × prior_mean + Σ ratings) ÷ (prior_weight + count)
New tools start at the prior mean — they earn their rank
Signal 03

Engagement Points

Beyond votes, we track meaningful engagement: website visits, comments, follows, and reviews. Each action contributes points to a tool's signal score, capped daily to prevent engagement farming.

Website click-through: +1 point (max 1 per tool/day)
Comment posted: +1 point to commenter
Product followed: +1 point (max 10/day per user)
Written review (30+ chars): +2 points to reviewer
All actions capped at 10 engagement points per user per day
Signal 04

Freshness Signal

An AI tool that hasn't been updated in months should not perpetually outrank an actively developed one. The freshness signal applies a time-decay multiplier to each tool's score based on recency of product activity.

Activity types: new version, changelog post, comment reply, new media
Active tools (updated within 7 days): 1.0× multiplier — no penalty
Dormant tools decay gradually to a floor multiplier of 0.4×
Returning from dormancy: freshness is restored on next update
Signal 05

Category Trending

Trend signals measure the velocity of votes and engagement within a category over a sliding time window — not just total volume. A category with 50 new votes in 24 hours ranks higher than one with 200 total votes but no recent activity.

Time window: rolling 24-hour and 7-day velocity buckets
Velocity score: (new votes in window) ÷ sqrt(total votes + 1)
Category signals update every 60 seconds via ISR
Used to surface "Rising" category badges on the Signals Hub

Data Privacy

Aggregated Only

All public signals on this hub show aggregated data. No individual user's voting history, engagement patterns, or personal identifiers are ever published or exposed.

No Individual Tracking Published

We collect engagement signals to power rankings, but we never sell individual data or expose per-user behavior publicly. Signal scores are computed server-side and published as aggregated category or tool-level metrics only.

Back to Signals Hub