Lead enrichment, one API, cheapest-first waterfall

agntdata Lead APIs sit on top of multiple upstream lead-data providers and expose a single unified surface per use case (email finder, email verifier, person/company enrich, person/company search, company discovery, domain emails, mobile finder, company intelligence). For each call you can pass an optional `fields[]` selector and an optional `max_cost_cents` budget — agntdata then waterfalls cheapest-first across providers that can fulfill those fields, stops as soon as the request is satisfied (or your budget runs out), and merges results into a normalized response shape. Provider names are never returned in the response. Pricing is hybrid: simple endpoints (e.g. email verifier, intelligence signals) are flat-priced; complex waterfalls (enrich, email finder, find-mobile, search) are billed in the published `min – max` cents range based on what was actually spent upstream, plus a small markup. Bring one API key, get one bill, and let agntdata handle vendor selection, key rotation, and result merging.

Which upstream providers are used?

agntdata abstracts upstream providers entirely. Provider names are never returned in the response and we routinely add or rotate providers as we improve coverage and price. You only ever talk to one API.

How is pricing calculated for waterfall endpoints?

Each waterfall endpoint publishes a `min – max` cents range. agntdata calls providers cheapest-first until your `fields[]` are filled (or your `max_cost_cents` budget is exhausted), then bills you the actual upstream cost plus a small markup. Flat-priced endpoints (email verifier, intelligence signals, etc.) are always charged at the published rate.

What does `max_cost_cents` do?

It caps the maximum amount agntdata will spend upstream on a single call. The waterfall stops as soon as the budget is reached, even if some requested fields are still missing. Lower values trade completeness for cost predictability.

How does field-based selection work?

Pass `fields[]` (e.g. `["email", "job_title"]`) and agntdata orders providers by which can fulfill those fields combined with cost. The waterfall stops as soon as every requested field has a value — saving you money compared to always running the full enrichment.

Are bulk endpoints faster than calling enrich repeatedly?

Yes — bulk endpoints fan out requests in parallel and aggregate billing into a single deduction. You also get per-record budget control via `max_cost_cents_per_record`.

Can I use these APIs from AI agents?

Yes — that's the primary use case. All endpoints return structured JSON optimized for function calling in GPT-4, Claude, Gemini, and similar models. You can also wire agntdata into MCP-based agent runtimes.