Tavily
Provides real-time web search and content extraction API for AI agents to access and reason over live web data.
67/100
- Entity0/25
- Machine-readable31.5/35
- Retrieval access17/20
- Footprint18.5/20
- Tavily67
- Data medianof 9 rated69
- Atlas medianof 537 rated61
- Retrieval botspartialverify
Whether the bots that decide citations are let in. Training bots are a separate list and are not counted here.
- llms.txtyesverify
A plain-text index at /llms.txt. Optional; read by some tools.
- Organization schemayes
schema.org Organization markup on the homepage, which is what a resolver reads a company's name and links from.
- Wikidata itemnone
Matched on the official website, never on the name. The substrate under the knowledge graph an engine resolves names against.
- Server-rendered words539swept
What a reader without JavaScript receives from the homepage.
- Identity score90/100swept
The analyzer's crawl-only section, 0-100: identity files, schema, crawler policy, sitemap. Not the analyzer score.
- Footprint5.1k GitHub stars
90/100
identity score
66
69
61
74
82
83
Not measured with the analyzer.
The full run asks four engines about this company by name. It runs on request, not on a schedule.
What is Tavily?
Provides real-time web search and content extraction API for AI agents to access and reason over live web data.
What is Tavily's Atlas Readiness?
67 of 100 on 2026-09-15, computed from four parts — entity, machine-readable, retrieval access, footprint — all four parts read. It places #6 of 9 rated Data tools.
Does Tavily let AI retrieval bots in?
Partly. On 2026-09-15, tavily.com fenced some paths from retrieval crawlers.
Does Tavily publish an llms.txt?
Yes: tavily.com/llms.txt was present on 2026-09-15.
Is Tavily a Wikidata entity?
No item whose official website is tavily.com was found on 2026-09-15.
How to read a row
- swept
- Read by a crawler: HTTP and parsing, no model call. Every row, every week.
- measured
- The full analyzer run, asked of four engines. Shown only where the company agreed to show it.
Can a retrieval engine find, resolve and quote this site? Four parts, each read by the sweep, each shown beside the total.
Entity 25 · Machine-readable 35 · Retrieval access 20 · Footprint 20
A part that was never read is left out and the total is rescaled to what was. It is never counted as zero.