Bulk files, REST API, batch append, MCP. Pick per dataset.
Delivery is part of the product. The same data can arrive as a Parquet file in your bucket, as a keyed endpoint inside your app, as a match against a file you send us, or as tools an AI agent can call.
Bulk files. Full datasets or filtered cuts in CSV, JSON, NDJSON or Parquet, pushed to your S3 bucket, pulled from our SFTP, or fetched from a signed link. Every delivery carries a manifest with row counts and checksums, and large files are chunked so a failed transfer never leaves you half a file. Daily signals land before 9am Eastern; monthly refreshes ship in the first business week. Snowflake, BigQuery and Databricks shares are available for teams that prefer a table to a file.
REST API. The license API is live and publicly documented at api.compcurve.com. Valuation, comparable sales, identity resolution and phone verification run as private evaluation endpoints; ask and we will enable a scoped key rather than pointing you at a docs page that does not exist yet. All of them use per-key scopes on geography and fields, rate limits and usage metering. Demo keys let a developer evaluate with masked contact fields before any contract; production keys return everything you licensed. CSV, JSON and NDJSON output on list endpoints so a script and a warehouse loader use the same call.
Batch append. Send us a file, get it back richer. Inputs can be addresses, names plus city, phones, emails, hashed emails, VINs or company domains. We return owner, verified mobile with line type, deliverable email, property, mortgage and equity, vehicle, demographics or firmographics, each row with a match tier. We run a free test slice first and report the match rate by field before invoicing. Files from a few thousand rows to tens of millions, delivered in days, with optional monthly re-append.
MCP server. For teams building with agents, the same data is available as Model Context Protocol tools: search licenses, resolve an owner from an address, pull comparable sales, check a phone, launch a batch append. It uses the same keys, scopes and metering as the REST API, so what an agent can see is exactly what your contract allows. Works with Claude Desktop, Cursor, ChatGPT and any MCP-capable client. This one is in development rather than deployed, so ask to be put on the list and we will tell you honestly where it stands.
How it runs
| 1 | Choose Bulk, API, batch append, MCP, or a mix per dataset. |
| 2 | Credentials Your bucket or our SFTP for files; scoped keys for the API and MCP. |
| 3 | Test A sample file, a demo key, or a free test slice for appends. |
| 4 | Run Scheduled deliveries with manifests, metered API and MCP calls, appends returned with match rates. |
Questions we get
Parquet?
Yes, and we recommend it for anything over a few million rows.
Is the MCP server live?
In early access with the license, identity, comps and phone tools. Ask for a key and we will enable it on your account.
Can an append run through the API instead of a file?
Yes. Single-record lookups run through the REST API or MCP; files run through batch append. Same matching, same rates.