Skip to main content
3 min read

The ten data sources

Ten places hold the truth about a person and a company. None of them holds all of it. abm.dev reads all ten and returns one answer — eighty-nine canonical fields and forty signals, each one cited, scored, and explained.

The ten

abm.dev draws on ten sources. Among them: LinkedIn for professional history, Companies House for UK registry filings, Perplexity for research, Tavily for web search, and Hunter for email. Five more sit behind them.

You do not pick the source. You ask a question; abm.dev decides where the answer lives.

HTTP
POST https://api.abm.dev/api/v2/enrichments
x-api-key: YOUR_API_KEY
Content-Type: application/json

{
  "targets": [
    { "type": "person", "email": "[email protected]" }
  ]
}

Each target also takes a LinkedIn URL or a company domain. Then read the result.

HTTP
GET https://api.abm.dev/api/v2/enrichments/{id}
x-api-key: YOUR_API_KEY

Note

Send your key as either x-api-key (used above) or Authorization: Bearer on POST /api/v2/enrichments. The full, live contract is at the API reference.

One response, ten origins

Ten sources go in. One response comes out: eighty-nine canonical fields — forty-three about the person, forty-six about the company — and forty signals.

No collage of raw payloads. No ten bills. abm.dev resolves the ten into a single shape, and the price is per enrichment, all sources included.

See Canonical fields for the full list.

Every value is accountable

A field without a trail is a guess. abm.dev does not ship guesses.

Each value carries three things:

  • A source. Which provider it came from.
  • A confidence. A number from zero to one. Higher means more of the ten sources agree.
  • A selection_reason. Why that value was chosen.

A value is cited or it is not returned. No silent fallbacks. No fabricated facts. When abm.dev does not know, it says so.

More on the scale in Confidence scores.

For agents

Agents connect over MCP. The server at https://mcp.abm.dev/mcp exposes abm.dev’s enrichment tools inside Claude.

In Claude: Settings → Connectors → Custom → paste the URL. The exact tool list is not documented here — see /for-agents.

In Cursor, Claude Code, or Windsurf, skip the connector. Hand the agent a prompt: read abm.dev/llms.txt, request a key, save it to .env, run a test enrichment, report back with citations. The agent does the wiring.

For scripts and terminals

No agent in the loop? Call the REST endpoints directly. A key in the x-api-key header, a JSON body, the two endpoints above. See Getting started and the CLI docs — treat any CLI specifics as the source of truth there, not here.

Real time and batch

For an agent loop that needs the answer as it arrives, abm.dev streams over SSE in real time.

For batch work, webhooks fire when a job finishes. The webhook contract — paths, headers, retries — is not documented here; see the jobs reference.

What it costs

Per enrichment. No subscription. Credits never expire. All ten sources included in every call. From about €0.29 per enrichment. Packs:

PackPrice
30€2.89
100€9.29
500€36.99 (best value)
2,000€119.99

The Playground is free. Launch credits come with the code LAUNCHCODES. Full pricing at /#pricing; sign up for a key.

Who runs it

abm.dev is operated by Foxley Farm Operations Ltd, UK Company No. 16392009, Oakham, United Kingdom. GDPR-compliant.

Read next