AI Tooling Detector Company AI Adoption Signals for Clay

Give it a domain. It tells you whether the company just talks about AI, actually runs AI tooling on its site, or charges money for AI.

Most AI detectors return a boolean, which is close to useless for targeting. "Published an llms.txt" and "bills $0.99 per AI resolution" are the same answer to a yes-or-no question and completely different sales situations. This actor returns a four-level ai_maturity tier with the evidence behind it, as one flat Clay-ready row per domain.

What's Inside

• Features
• The ai_maturity ladder
• Input
• Output
• Pricing
• Usage Examples
• Error Handling
• Limitations

Features

• Four-level maturity tier, not a boolean. none, declared, deployed, commercialized. Each rung needs its own class of evidence.
• Pricing-page commercialization detection. Finds AI credits, token allowances, AI add-ons, AI-named plans, and per-outcome AI pricing. This is the tier that tells you a company already pays for AI.
• 50+ AI vendor fingerprints. Support agents (Sierra, Decagon, Ada, Intercom Fin, Forethought), AI search (Glean, Algolia, Inkeep, Kapa), personalization (Mutiny, Dynamic Yield), direct inference endpoints (OpenAI, Anthropic, Azure OpenAI, Bedrock, Gemini, Groq, Together, Fireworks), vector stores, and app frameworks.
• Validated llms.txt, not just a 200. A lot of single-page apps answer /llms.txt with their HTML shell. Those return has_llms_txt: false with a reject reason, so the strongest signal in the pack stays clean.
• robots.txt AI crawler policy. Reports which of 26 AI crawlers a company allows or blocks, with the rule that produced the verdict.
• An explicit blocked flag. A domain behind a bot challenge is reported as blocked at low confidence, never as a confident no. Without it a challenged domain looks identical to a company with no AI.
• Evidence you can quote. Every verdict ships the strings it was built from, so a rep can paste one into an email.
• Batch input. One domain or an array of them, analyzed concurrently.
• HTTP only. No browser, no proxy, 256MB. Runs finish in seconds.

The ai_maturity ladder

Tier	What it means	What proves it
commercialized	The company charges for AI	AI credits, token allowances, an AI add-on, an AI-named plan, or a per-outcome AI price on the pricing page
deployed	AI tooling is running on the site	An AI-native vendor, a direct inference endpoint, or an AI-capable platform backed by AI copy
declared	The company says AI, nothing observable	A validated llms.txt, AI crawler rules in robots.txt, or weighted marketing copy
none	No signal fired	Nothing

The ladder is strict. An llms.txt can never on its own reach deployed, and a chat widget can never on its own reach commercialized.

Input

Field	Type	Default	Description
domain	string	intercom.com	A single company domain. Ignored when domains is filled.
domains	array	[]	Batch mode. Takes precedence over domain.
check_pricing	boolean	true	Fetch and score the pricing page. Turning this off makes runs faster but caps results at deployed, since commercialized can only be proven on a pricing page.
batchSize	integer	5	Domains analyzed concurrently, 1 to 10.
request_timeout_ms	integer	9000	Per-request timeout, 3000 to 20000.
skipCache	boolean	false	Ignore the 7 day result cache.
qa_test_mode	boolean	false	Emit a single summary row and skip all network work.

Output

One flat row per domain, snake_case, Clay-native.

Field	Type	Example
domain	string	intercom.com
resolved_url	string	https://www.intercom.com/
http_status	integer	200
ai_maturity	string	commercialized
uses_ai	boolean	true
confidence	number	1
has_llms_txt	boolean	true
llms_txt_url	string	https://intercom.com/llms.txt
llms_txt_reject_reason	string	html_catch_all_route
has_ai_txt	boolean	false
robots_ai_directives	array	[{"agent":"GPTBot","policy":"block","rule":"disallow: /"}]
detected_tools	array	[{"vendor":"Intercom Fin","category":"support","ai_native":true,"confidence":"high","evidence":"markup match: fin.ai"}]
detected_tool_names	string	Intercom Fin
detected_categories	array	["support"]
ai_copy_matches	array	[{"pattern":"ai_assistant","weight":2,"evidence":"AI Agent"}]
ai_copy_weight	integer	8
ai_pricing_detected	boolean	true
ai_pricing_score	integer	10
pricing_url	string	https://www.intercom.com/pricing
evidence	array	["AI priced per resolved outcome: \"$0.99 per Fin outcome\""]
signals_fired	array	["llms_txt","vendor_fingerprint","pricing_commercialization"]
signals_missing	array	["ai_txt","robots_ai_directives"]
blocked	boolean	false
blocked_reason	string	http_202_challenge
is_js_rendered	boolean	false
fetch_error	string	null
is_summary_row	boolean	false
notice	string	null
checked_at	string	2026-08-05T17:20:00.000Z

Read confidence and signals_missing before trusting a none. A verdict built on one reachable page should not be read like a verdict built on five.

Pricing

Pay per event. You are charged once per domain analyzed.

Tier	Per domain	Per 1,000 domains
Free	$0.003	$3.00
Bronze	$0.00285	$2.85
Silver	$0.0027	$2.70
Gold	$0.00255	$2.55

Actor start is $0.00005 per run.

Free Apify-plan users get 15 results per calendar month. Paid plans are unlimited. When the free cap is reached the run exits cleanly with a notice row rather than an error.

Repeat lookups inside 7 days are served from cache.

Usage Examples

Apify Console / API

curl -X POST "https://api.apify.com/v2/acts/EwkHhmqiuJgRoVEbE/runs?token=YOUR_TOKEN" \
  -H "content-type: application/json" \
  -d '{
    "domains": ["intercom.com", "notion.so", "berkshirehathaway.com"],
    "check_pricing": true,
    "batchSize": 5
  }'

Clay Integration

Add an Apify enrichment column, point it at this actor, and map your domain column to domain. Every output field lands as its own Clay column.

Useful follow-on formulas:

• Target companies that already buy AI: ai_maturity = "commercialized"
• Target companies that talk about AI but have not shipped it: ai_maturity = "declared"
• Suppress unreliable rows: blocked = false AND confidence >= 0.6

MCP Integration

{
  "mcpServers": {
    "mamba-ai-tooling-detector": {
      "command": "npx",
      "args": ["-y", "@mambalabsdev/mcp-ai-tooling-detector"],
      "env": { "APIFY_TOKEN": "your_apify_token" }
    }
  }
}

This actor is also a tool on the Mamba Labs GTM Suite server, which exposes the whole fleet through one install.

Error Handling

Condition	Behavior	Output
Homepage returns an anti-bot challenge	Row is still emitted, flagged, and confidence pinned to 0.15	blocked: true, blocked_reason set
Homepage returns 200 with no server-side text	Row emitted and flagged as a client-rendered shell	is_js_rendered: true, confidence reduced
Pricing page unreachable or missing	Analysis continues; the verdict caps at deployed	pricing_url: null, ai_pricing_score: 0
/llms.txt answers 200 with an HTML shell	Not counted as an llms.txt	has_llms_txt: false, llms_txt_reject_reason set
Connection reset or timeout	Row emitted with the transport error recorded	fetch_error set, http_status: null
Key-value store unavailable	Fails open: runs without the cache and without quota enforcement	Normal rows
Free-tier monthly cap reached	Exits cleanly before any network work	One notice row, is_summary_row: true
No domain supplied	Exits cleanly	One notice row

The run always produces at least one dataset item.

Limitations

• Vendor recall is capped by what the server sends. Modern chat widgets are injected by client-side JavaScript, so a company running an AI widget that never appears in the served markup will read as declared rather than deployed. On a 15 domain AI-forward sample, 9 reached deployed or better and 14 reached at least declared. Adding a browser would raise recall and would also raise the memory floor to 1024MB and the price with it. The tradeoff was made deliberately in favor of price.
• Anti-bot challenges affect roughly 10% of domains, concentrated in large consumer brands and retail. Those rows come back blocked: true rather than silently negative. No proxy is configured by default.
• commercialized requires a reachable pricing page. Companies that gate pricing behind "contact sales" cannot be proven to commercialize AI from the outside.
• declared is a claim, not a deployment. That is the point of separating it, but do not read it as adoption.
• llms.txt prevalence is only 5 to 13%. It is a high-precision positive whose absence proves nothing.
• Fingerprints need maintenance. Vendors change script URLs. The category's incumbent has gone 138 days without a rebuild, which is what fingerprint rot looks like.

---

Part of the Mamba Labs GTM Intelligence Suite

Actor	Actor ID
GTM Hiring Signal Scraper	D7O1SA2EqwHGsGr1P
GTM Tech Stack Signal Enrichment	qyd7nNyqFPelQViBx
GTM Signals Aggregator	xKdRfnfFNkdMpFuNs
Job Board Keyword Signal Scanner	4DvqpvhMR74NLcDDY
Domain to LinkedIn URL Resolver	3HtnSaqPHOg1Qg5gx
ICP Fit Scorer	W161DT8W4kW55dMFh
Domain Deliverability Checker	0tVgxI7A6o9jMlxmc
Company Firmographic Enricher	YlUtLWjfPpqykmB8g
Company Social Presence Mapper	4k6CCemkgBDz18m2h
Company Identity Resolver	lr8fTRAmZCBZmuwwh
Funding & Press Signal Scanner	FS13X6dhQVgX3XOM6
Company Change-Event Feed	oX44rS0fkEJ3rXLWe
Job Posting Monitor	QCfICD9kOPiOdj5iI
AI Tooling Detector	EwkHhmqiuJgRoVEbE

Built by Mamba Labs.