Datasets
Structured data, with a public contract.
Typed records behind a schema you publish, served from a fast public URL your own site and your agents can read — without a deploy, and without an API key.
Data at rest, not work in flight.
A board is where work moves: columns, review, approval. That is the wrong shape for a list of journal posts, a table of room rates, or the settings your website reads at request time. Those do not move through states — they sit still and get read, constantly, by something that is not a person.
A dataset is that second thing. Each one has a typed schema with stable machine keys, so renaming a label never breaks a consumer and an invalid record never quietly passes through. Publish a record and it appears on a cached delivery URL immediately.
There are two kinds. A collection holds many records. A document is a schema-validated singleton for configuration — which is what makes changing a booking window an edit rather than a pull request.
Inside Datasets
Typed schemas
Text, number, boolean, date, select, URL, email, JSON, media, and references. Every field has an immutable machine key, so labels rename freely.
Cached delivery
Published records are served from a token-authed public URL with real cache headers — no API key, and no function invocation per read.
Documents as config
A validated singleton for the settings a site reads at request time. Change one and it takes effect now.
Versioned schemas
Every schema change is recorded with its full field snapshot, so what changed between two versions is a lookup rather than a replay.
Scoped writes
Reading, writing records, and changing the schema are three separate scopes — a record-writing agent key cannot mutate the contract.
Machine-operable
The full management surface is on the REST API and MCP, so an agent can create a dataset, define its schema, and publish to it.
Publish it once, read it everywhere.
Book a 30-minute call. We'll go through what's working, what isn't, and what the first move should be. If we're not the right fit, we'll say so on the call.