About
How Bench's AI tool ratings works.
A deterministic, multi-source rating system for AI developer tools — built by WWT to keep our engineering and client-advisory practices grounded in evidence.
PILLAR 01
Review platforms
G2, Capterra, TrustRadius, Product Hunt, GetApp. Star ratings normalized with a 9-month recency half-life so old reviews still count, but less.
PILLAR 02
Community sentiment
Vendor forums, GitHub Issues + Discussions, Hacker News. Threads scored per dimension by a deterministic LLM pass with cached prompts.
PILLAR 03
Practitioner ratings
Engineers rate tools directly through the site. Submissions are credibility-weighted from GitHub profile, ecosystem activity, and WWT employment.
What you see
Six dimensions, one radar.
Each tool gets a 0–100 composite and six per-dimension subscores. The fixed axis order means any two tools' radars overlay directly — no mental remapping. Click into a dimension on the tool detail page to see every contributing signal with citation links.