How we score AI workflow tools
The launch methodology weights workflow fit, setup friction, orchestration potential, usability/documentation confidence, pricing clarity, and maturity signals.
- Workflow fit: 30%
- Setup friction / switching cost: 20%
- Integration and orchestration potential: 20%
- Usability and documentation confidence: 15%
- Commercial clarity: 10%
- Support and maturity signals: 5%
When we use scores and when we do not
We avoid fake precision. Numeric scores appear only when evidence is consistent enough across a subcategory. Otherwise we publish a qualitative verdict and explicit uncertainty.
- Researched pages can still rank tools, but must not imply hands-on testing
- If evidence is thin, we downgrade labels or keep the page in draft
- Affiliate availability never overrides the editorial score
Evidence labels
Every live page is labeled as Researched, Verified docs, Field notes, or Data backed. Tested and Expert reviewed are reserved internal states until real evidence or reviewer records exist.
- Researched is the default for desk-research pages
- Verified docs means the page was checked against official docs, pricing, integration, or support materials
- Field notes means operator observations were documented from a real browser or runtime workflow
- Data backed requires a named source and explicit method