This is the simplest practical API Builder journey: choose the namespace, define the product, connect the data, secure access, and test the result before operational handoff.
Example sequence
- Open API Builder.
- Choose the workspace namespace and published domain.
- Choose the API type.
- Attach the source dataset or datasets.
- Configure request matching and response fields.
- Define access controls.
- Save and index the product.
- Test the API before you hand it to another team or client.
- Use Logs, Monitoring, and Traffic once requests begin flowing.
How API Builder reads your data
API Builder works with managed GeoParquet-backed vector layers only. Upload ingest, empty create, and future enterprise connectors all use the same on-platform contract: a GeoParquet managed snapshot plus hot edit deltas in Postgres during active edit sessions.
When you save or rebuild an API product, Tamaat:
- Reads the bound layer/version row through DuckDB (
GeoParquet snapshot + hot deltas). - Materializes a search index in Postgres (
api_product_documents). - Serves public autocomplete, geocode, and query requests from that materialized index for low latency.
Index consistency: The materialized index reflects the GeoParquet and hot-delta state observed while the rebuild queries run. It is not a transactionally frozen snapshot across the entire layer. If the source layer changes during rebuild, the index converges on the next rebuild after the product is marked stale.
Stale index: While a rebuild is in progress, public endpoints may return 503 Service Unavailable with a retry hint rather than silently serving outdated results.