AI Ark

Give your agents AI Ark’s B2B database — 400M+ people and 70M+ company profiles, verified emails and mobile numbers, lookalike discovery, reverse lookup, and personality analysis — through agntdata.

Overview

The AI Ark connection lets deployed agents build outbound lead lists and enrich contacts from your AI Ark account: search people (400M+ profiles) and companies (70M+ profiles, including lookalike search from up to 5 seed companies), preview a market for free-ish sizing before paying for full results, find verified emails and mobile numbers, run bulk export jobs (search + emails in one async job, up to 10,000 people, optional webhook delivery), reverse-lookup a full profile from an email address, and generate a DISC/OCEAN personality analysis with outreach guidance from a LinkedIn URL.

• Base URL: https://api.agntdata.dev/v1/connections/ai-ark
• AI Ark auth: an AI Ark API key. Paste it once in the dashboard — agntdata stores it encrypted and attaches it (as AI Ark’s X-TOKEN header) to every call, so you never send it yourself.
• MCP tools: agents call connection_ai_ark_* tools, e.g. connection_ai_ark_people_search, connection_ai_ark_company_search, connection_ai_ark_people_email_finder_by_track_id, connection_ai_ark_fetch_credit.

Usage spends AI Ark credits from your AI Ark account. You pay only for what comes back: 0.5 credits per person search result, 0.1 per company result, 1 per found valid email, 5 per found mobile number, 4 per personality analysis. Previews (1 credit per page), lists, statistics, submissions, and the credit-balance check are free or near-free, and enrichment that finds nothing costs 0. Check the balance anytime with connection_ai_ark_fetch_credit.

Search → email flow uses a trackId. People Search returns full profiles without contact data plus a trackId; pass that to Find Emails by Track ID within 6 hours (single use) to get verified emails for that exact result set. Email finding is asynchronous — poll the statistics endpoint until state: DONE, then fetch the results, or register a webhook.

Setup

1. Create an AI Ark API key

   In AI Ark open Settings → API Management (direct link) and create a key.

2. Connect AI Ark

   In the agntdata dashboard open Integrations, pick AI Ark, and paste the key.

3. Pick AI Ark tools

   In an agent’s Tools panel, add the tools the agent needs — for example people_search, company_search, people_email_finder_by_track_id, and fetch_credit. Search and preview tools run automatically; enrichment tools that spend meaningful credits (email finder, exports, mobile phone finder, personality analysis) ask for approval by default, and you can tune that per agent.

Examples

Check your credit balance

curl "https://api.agntdata.dev/v1/connections/ai-ark/v1/payments/credits" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY"

Search companies (German software, 51–200 employees)

curl -X POST "https://api.agntdata.dev/v1/connections/ai-ark/v1/companies" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "account": {
      "industries": { "any": { "include": { "mode": "SMART", "content": ["software development"] } } },
      "location": { "any": { "include": ["Germany"] } },
      "employeeSize": { "type": "RANGE", "range": [{ "start": 51, "end": 200 }] }
    },
    "page": 0,
    "size": 10
  }'

Search people, then find their emails

# 1. Search — returns profiles plus a single-use trackId (expires in 6 hours)
curl -X POST "https://api.agntdata.dev/v1/connections/ai-ark/v1/people" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contact": { "jobTitle": { "any": { "include": { "mode": "SMART", "content": ["head of sales"] } } } },
    "account": { "location": { "any": { "include": ["United States"] } } },
    "page": 0,
    "size": 10
  }'

# 2. Submit the async email-finder job with the trackId from step 1
curl -X POST "https://api.agntdata.dev/v1/connections/ai-ark/v1/people/email-finder" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "trackId": "TRACK_ID_FROM_SEARCH" }'

# 3. Poll status, then fetch results when state is DONE
curl "https://api.agntdata.dev/v1/connections/ai-ark/v1/people/email-finder/$TRACK_ID/statistics" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY"
curl "https://api.agntdata.dev/v1/connections/ai-ark/v1/people/email-finder/$TRACK_ID/inquiries?page=0&size=100" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY"

Reverse-lookup a profile from an email

curl -X POST "https://api.agntdata.dev/v1/connections/ai-ark/v1/people/reverse-lookup" \
  -H "Authorization: Bearer $AGNTDATA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "email": "[e-mail]" }'

Notes

• page and size are required on the search endpoints (page is zero-based, size max 100; People Preview: 25–100). Pagination retrieves up to 10,000 records per search; to go beyond, exclude already-fetched IDs with a List (POST /v1/lists, free, expires after 24 h).
• Text filters take a mode: SMART (AI-related concepts), WORD (phrase match), STRICT (exact). any = OR, all = AND; both accept include/exclude; max 300 values per array.
• No matches returns an empty content array or 404 data not found — treat both as “no results”, not an error.
• Rate limit: 5 requests/second per key (429 on excess). Async jobs allow up to 500 pending submissions per service; charged-but-undelivered jobs auto-refund within 10 hours.