Agencies by the tools they work with
What a firm can measure decides what it can prove to you. An agency selling AI search work that cannot name how it tracks appearance in AI answers is not disqualified — but it is a question worth asking before you sign anything.
We are still establishing stacks for the firms we list. Declarations appear here as we verify them.
AI search visibility
Tracks whether a brand appears in AI answers. The category barely existed two years ago, and a firm that cannot name one of these cannot show you what it moved.
- Profound —
- Peec AI —
- Otterly.AI —
- Scrunch AI —
- Ahrefs Brand Radar —
- Semrush AI Toolkit —
- Goodie AI —
Models and assistants
The models the firm builds on. Naming one is table stakes; naming which one for which job is the useful answer.
- ChatGPT / OpenAI models —
- Claude —
- Gemini —
- Perplexity —
- Llama (self-hosted) —
SEO platforms and crawlers
Classic SEO tooling. AI search sits on top of technical SEO, so a firm with no crawler in its stack is telling you something.
- Ahrefs —
- Semrush —
- Screaming Frog —
- Sitebulb —
- Botify —
- Conductor —
- Google Search Console —
Content and briefing
Where the writing gets planned and scored. Worth knowing if you care whether output is templated.
- Clearscope —
- MarketMuse —
- Surfer SEO —
- Frase —
Automation and orchestration
How work gets wired together. This is usually where an agency's actual leverage lives, and where you find out whether you own the workflows afterwards.
- n8n —
- Zapier —
- Make —
- LangChain —
- LlamaIndex —
Analytics and reporting
Where the numbers you get sent come from.
- Google Analytics 4 —
- Looker Studio —
- BigQuery —
Why we collect this
Directories usually render tools as a tag cloud the agency fills in itself, which makes the field free to inflate and worthless to search. Ours works differently in two ways.
- Every declaration carries how we established it — verified in public record, reviewed privately, or simply declared by the firm. You can see which.
- Claiming a tool cannot move a firm up any list. The scorer has no access to this data at all, and the deploy gate asserts the ordering is identical with and without it.