Babel

Waitlist open · launching soon

The Schema Control PlaneWe All Waited For

One source of truth for your schemas, versioned like code, and every system in your pipeline kept in step with it.

  • Open source
  • Confluent-compatible
  • Object-store native
// capabilities

Compatible where it counts.
Different where it matters.

version_control

Git for your schemas

Branch a subject, evolve it independently, and merge it back when it is compatible. Roll back by appending, so history is never rewritten.

point_in_time

Point-in-time reads

Ask what any subject looked like at any instant. Every version, fork, and merge is an immutable record you can audit.

stateless

Stateless architecture

No Kafka as a backend. Point it at any globally available object store and it scales horizontally, with no leader and no coordination service.

drop_in

Drop-in compatible

API-compatible with the Confluent Schema Registry. Point your existing clients at Babel with no code changes, and import your Confluent, Aiven, or Redpanda registry in place.

approvals

Changes get approved

Route a proposed change to the people who own the schema, in Slack, and hold it until they approve. Every decision is recorded against the version.

agents

Built for agents

An MCP server exposes subjects, versions, lineage, and compatibility checks. An agent proposes a change on a fork, a human approves the merge.

everywhere

One truth, every representation

The same schema lives in your database, your streams, your lake, and your warehouse, and they drift apart. Babel watches all of them and keeps every representation in agreement with the version you approved. This is the part a registry never did.