Embedding Sync
- Postgres, S3 and Kafka sources
- Bring your own model or ours
- Searchable seconds after a write
- Backfills without downtime
Point Nodeform at the table, bucket or topic where your content lives. New rows are embedded, indexed and searchable within seconds, and re-embedding a model change is one command.
The pipeline you were going to write anyway
Every team that adopts vector search ends up building the same job: watch a source, embed the new rows, upsert them, handle failures, backfill after a model change. Nodeform runs it, with dead-letter handling and a backfill that never takes the index offline.
Why sync belongs in the platform
Freshness you can state
p95 from write to searchable is under four seconds on the Scale tier, and it is on the same dashboard as query latency.
Model changes are routine
Re-embed a collection against a new model on a shadow index, compare recall, then cut over. The old index serves until you are happy.
Your model, not ours
Send vectors you computed, or give us an endpoint. Nodeform stores and searches; it does not decide what your embeddings mean.
Ready to stop running
your own vector index?
Read the quickstart, or talk to the engineer who will run your index.
