Enrich
Give what you know — a name and a domain, an email, a profile URL — and get back what your configured providers can add, with the source of each field attached.
Loading
Everything the MCP surface does is a plain HTTP call underneath — for pipelines, back-office jobs and anything that is not an assistant.
The same capability, for systems that are not agents. Everything the MCP surface can do is a plain HTTP call underneath, because an agent-only product is a product that cannot be put in a cron job.
Give what you know — a name and a domain, an email, a profile URL — and get back what your configured providers can add, with the source of each field attached.
Declare provider order, per-provider budgets and the stopping rule. The expensive provider only gets asked when the cheap one has already failed.
Deliverability and validity checking as a separate step, so an unverified record never reaches a sending system and quietly damages a domain.
Submit a batch, get called back when it finishes. Long jobs should not require you to hold a connection open.
A queryable record of every lookup: which provider, what it cost, cache hit or miss, and why the waterfall stopped.
These are not gaps to be filled later. They are the shape of the thing, and if any of them is a dealbreaker for you then an all-in-one vendor is genuinely the better choice.
Out of scope, on purpose
If you would use this, tell us which providers you already pay for and what you would want the first endpoint to do. Early access decides the build order.