← Back

Methodology

Data Collection

Each tool's data is collected automatically using the following process:

  1. Fetch official pages (website, documentation, GitHub, pricing, privacy policy)
  2. Use LLM to extract structured metadata (platform support, features, compliance)
  3. Compare with current data in our database
  4. Store differences in a staging table for validation

3-Day Confirmation Rule

To prevent false positives from temporary website changes or scraping errors, we require that the same data be detected on 3 consecutive days before it appears on this site:

  • Day 1: Change detected → stored in staging
  • Day 2: Same change confirmed → counter incremented
  • Day 3: Same change confirmed again → promoted to production

This ensures data integrity and reduces noise from temporary website updates.

Feature Definitions

Model Choice
User can select between multiple LLM providers (e.g., Claude vs GPT vs Mistral)
Freemium
Free tier available with limited features
Enterprise Offer
Dedicated support, SLA, or enterprise licensing available
Local LLM
Can run with local or self-hosted LLM models
Computer Use
Can control your computer directly (click, type, etc.)
Sandboxing
Actions are sandboxed or limited to prevent system damage

Compliance Flags

Compliance flags indicate whether a tool claims compliance with or adherence to these regulations:

  • GDPR: General Data Protection Regulation (EU privacy)
  • AI Act: EU Artificial Intelligence Act compliance
  • DSA: Digital Services Act (EU platform regulation)
  • DMA: Digital Markets Act (EU competition law)
  • Cyber Resilience: Cyber Resilience Act compliance
  • SOC 2: Service Organization Control (security audit)

Limitations

  • Data reflects official claims; we don't independently verify functionality
  • Public websites only; private APIs and credentials not supported
  • Compliance flags are based on public statements, not audits
  • Platform support limited to desktop (macOS, Windows, Linux) and web interface