Vendors and Technologies
List the technologies detected across 22 million companies, find the companies running one, and filter people by their employer's stack.
GET /leads/vendors · GET /leads/vendors/{slug}/companies · GET /leads/companies/{id}
The technology a company buys is the strongest buying signal in the dataset. A company running a competitor's product is in the market for the category; one running a complement is already set up to adopt. These endpoints turn that into a list.
What detection actually is
Each detection records where it was seen and what matched, so every claim can be audited rather than taken on trust:
| Field | Meaning |
|---|---|
slug | The stable identifier. This is also the value the vendors profile facet accepts |
name | Display name, e.g. Microsoft Exchange Online |
confidence | verified or shared — see below |
via | The evidence type: host, mx, txt |
matched | The literal string that matched, e.g. static.cloudflareinsights.com |
confidence is not a score, and the two values mean different things:
verified— the evidence identifies this vendor and only this vendor. Anazureedge.netCDN host, adynatrace-site-verificationTXT record.shared— a real signal used by many products. Agoogle-site-verificationrecord proves Search Console and nothing more; reading it as "runs Google Cloud" would be wrong.
An absent vendors array means not scanned, not "uses nothing". Detection has
covered roughly 36% of the 22.3 million companies so far. vendorsDetectedAt is the
field that separates the two: a company with no vendors and no timestamp has not been
looked at, while one with a timestamp and an empty array genuinely matched nothing.
Treating the first as a zero is the most common way to misread this data.
GET /leads/vendors
Every detected technology, most companies first.
| Parameter | Notes |
|---|---|
limit | Default and maximum 200 |
sample | How many companies the tally reads. Default 50,000, maximum 500,000 |
curl -u "$LEADS_USER:$LEADS_PASS" "https://api.connectsafely.ai/leads/vendors"{
"results": [
{ "slug": "google-tag-manager", "name": "Google Tag Manager", "companies": 1264, "verified": 1264 },
{ "slug": "google-search-console", "name": "Google Search Console", "companies": 1079, "verified": 0 },
{ "slug": "cloudflare", "name": "Cloudflare", "companies": 480, "verified": 480 }
],
"count": 3,
"sample": 50000,
"scanned": 9990,
"estimated": true
}estimated is always true, and sample/scanned say what the counts came from. An
exact tally means unwinding the vendors array across every scanned company — minutes of
work to rank a dropdown — so the counts are a bounded sample and the response is explicit
about it. Note google-search-console: 1,079 companies, 0 verified, because its only
evidence is a shared signal.
The ranking is computed at most once every 15 minutes and served from memory after
that. It is an estimate over a bounded slice either way — estimated: true and sample
say so — and the slice does not change between requests, so recomputing it per call only
made the endpoint slow: 1.2-1.9s at rest, and 504 under load.
Pass a different ?sample= to get a different estimate; each size is cached separately.
Note the ranking is not stable across sample sizes: 25,000 against 50,000 agrees on
only 4 of the top 25 positions, so compare numbers only within one sample size.
GET /leads/vendors/{slug}/companies
The companies detected running one technology, largest first.
| Parameter | Notes |
|---|---|
confidence | verified or shared. Constrains the same detection the slug matched |
min_employees | Drop companies below a headcount |
limit | Default 25, maximum 200 |
industry company_type revenue tech_categories specialty | Company facets, which narrow the vendor's companies |
The slug in the path is the vendors filter, so ?vendors= is a 400 here rather
than a second filter that could disagree with it. Everything else in the
company facet table composes:
# Nonprofits running Cloudflare, 500 people or more
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/vendors/cloudflare/companies?company_type=Nonprofit&min_employees=500"Note that industry, company_type and specialty do not split on comma — their
own values contain one. Repeat the parameter to OR.
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/vendors/cloudflare/companies?confidence=verified&min_employees=500"{
"vendor": "cloudflare",
"confidence": "verified",
"results": [
{
"id": "6a503c12cb31b206160294d8",
"name": "Bloomberg",
"domains": ["bloomberg.com"],
"employeeCount": 26777,
"industry": "Financial Services",
"headquarters": "New York, NY",
"vendorCount": 14
}
],
"count": 1,
"hasMore": true
}confidence filters the detection that matched the slug, not the company. Without that
distinction, "Cloudflare at verified confidence" would also return a company that carries
Cloudflare as a shared signal and something else as verified.
GET /leads/companies/{id}
One company in full: description, headcount, revenue, specialties, and the complete
vendor list with its evidence. The id is the id returned by
GET /leads/companies and by the vendor endpoint above.
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/companies/6a503c12cb31b206160294d8"{
"id": "6a503c12cb31b206160294d8",
"company": {
"name": "Bloomberg",
"domains": ["bloomberg.com"],
"websites": ["https://www.bloomberg.com"],
"linkedInUrl": "https://www.linkedin.com/company/bloomberg",
"industryLinkedIn": "Financial Services",
"employeeCount": 26777,
"companySize": "10,001+",
"revenueRange": "$1B+",
"followers": 2254330,
"headquarters": "New York, NY",
"vendors": [
{
"slug": "cloudflare",
"name": "Cloudflare",
"confidence": "verified",
"via": ["host"],
"matched": ["static.cloudflareinsights.com"]
}
],
"vendorCount": 14,
"vendorCountVerified": 13,
"vendorsDetectedAt": "2026-10-09T17:16:35.660Z"
}
}A malformed identifier is a 400 invalid_id, and an identifier that matches nothing is a
404 not_found — different problems, so they are reported differently.
From a technology to the people
The slug values above are the same vocabulary the vendors profile facet accepts, so
the two halves compose: pick a technology, then ask for the decision-makers at the
companies running it.
# Who runs Cloudflare?
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/vendors/cloudflare/companies"
# Engineering leaders in the US at companies running Cloudflare, with a work email
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/search?country_code=US&job_function=Engineering&vendors=cloudflare&has_work_email=true"The profile facet carries the company's stack, not the individual's, which is what makes the second query mean "people at companies running Cloudflare".
The same route, through company ids
The profile copy of the company's stack is going away: a profile will carry companyId
and nothing else about its employer. The durable route is two calls — narrow companies,
then pass their ids to search:
# 1. The companies, filtered however you like on the company itself.
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/vendors/cloudflare/companies?company_type=Public%20Company&limit=50"
# 2. The people at them. Repeat company_id for each id from step 1.
curl -u "$LEADS_USER:$LEADS_PASS" \
"https://api.connectsafely.ai/leads/search?company_id=6a503c12cb31b206160294d8&company_id=6a503c1acb31b2061605293d&job_level=VP"company_id has its own index, several ids OR, and the count is uncapped — 1,000 ids
answer in 55ms on the live collection. It is also the more accurate question: it matches
the company a profile resolves to rather than the string it happens to carry.
See the Facets Reference for the rest of the surface.
Describe Facets
Return every filterable parameter on the leads database with its type, countability and the servable index combinations, as JSON.
Enrich a Person
Resolve one email address or LinkedIn URL against the leads database and return the full profile and investor record. Makes no external call and consumes no credits.
