Most data does not start its life in Tinybird. It lives in a Kafka topic, an S3 bucket, a Postgres database, or a data warehouse. A connector wraps the movement from that system into a Tinybird data source: it handles authentication, schema mapping, incremental sync, retries, and monitoring, so a stream of rows lands in the right columnar table with the right types.

Connectors split into two shapes. Streaming connectors (Kafka, change-data-capture from Postgres or MySQL) keep a continuous tail of new rows flowing into a data source as they appear at the source. Scheduled connectors (S3, BigQuery, Snowflake) run on a cadence and copy new partitions or files since the last run.

Connectors close the gap between existing data infrastructure and the engine underneath. They replace the bespoke ingest service that would otherwise sit between a queue or warehouse and the analytics layer.