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.
User guide
The dashboard sorts every tool by composite score. Narrow with the category chips, then click into a tool to see its radar, dimension breakdown and the per-source signals behind each number — every value links back to its citation.
Rating a tool
- 1Sign in with GitHub
First-time sign-ins create your rater profile automatically — no separate signup.
- 2Open any tool, hit “Rate this tool”
The button sits in the header of every tool detail page.
- 3Score each dimension — or skip
Score 1–10, or hit Skip. Skipping is honest data; forcing a guess hurts score quality.
- 4Your weight reflects credibility
Rating weight is derived from your GitHub profile and ecosystem activity. WWT engineers show a verified-expert badge.
- 5Latest submission wins
Re-rating supersedes your earlier scores for the same dimension — no duplicate counting.
Comparing & suggesting
Pick up to four tools on the compare page — the fixed axis order means any two radars overlay directly, and the table underneath gives exact per-dimension numbers with confidence. Spot a tool we don't track yet? Suggest it: name, URL and category land in a moderation queue, and once approved the tool appears with no scores until the next ingestion run picks it up.