Run SQL, migrate data, and keep systems in sync — all in one tool.
One query. Multiple sources.
SELECT *
FROM read_parquet('orders.parquet') o
JOIN postgres.public.customers c
ON o.customer_id = c.id
LIMIT 10;SELECT *
FROM read_parquet('orders.parquet') o
JOIN postgres.public.customers c
ON o.customer_id = c.id
LIMIT 10;Join a Parquet file with a live PostgreSQL table — then use the same query as a stream source. No export. No rebuild. No second tool.
DBConvert Streams gives you one self-hosted product for working with databases end to end: a free IDE for browsing schemas, running SQL across PostgreSQL, MySQL, files, and S3, and editing data inline; and a paid streaming engine for bulk migration with automatic schema conversion and log-based CDC. Same install, same UI, same API.
The point is to remove the gap between "I need to look at this data" and "I need to move this data." You explore, validate, and configure in the IDE — for free — and only pay when you actually run a migration or replication job.
Explore data, run migrations, and keep target systems in sync.
Now with AI
You: Is my Parquet export of orders still in sync with the live Postgres table?
AI: Joined both in one cross-source query, matched on order_id: both hold 5,000 rows, 0 missing on either side. The snapshot is still current.
You: Is my Parquet export of orders still in sync with the live Postgres table?
AI: Joined both in one cross-source query, matched on order_id: both hold 5,000 rows, 0 missing on either side. The snapshot is still current.
Connect Claude, Cursor, or Copilot via MCP — your assistant reads live connections, schemas, and streams. Read-only, on your infrastructure.
Explore AI AssistantsFiles and object storage are first-class query sources, not external imports.
Cross-source SQL makes validation and data prep faster than switching between tools.
Use one interface for SQL exploration, migration execution, CDC status, and run history.
Use databases as sources and targets. Treat files and S3 as queryable datasets.
More connectors added regularly.
Your laptop, a VPS, or your own server. No cloud dependency, no vendor lock-in.
Self-host on any machine — local server, VPS, or cloud VM.
1-click images for AWS, Azure, GCP, DigitalOcean, Hetzner, and Vultr.
A self-hosted product that combines a free SQL IDE for PostgreSQL, MySQL, files, and S3 with a paid streaming engine for one-time migration and ongoing CDC replication. One install handles exploration, validation, migration, and replication.
Yes. The IDE — connections, schema browsing, SQL console, cross-database queries, ER diagrams, file and S3 access, inline editing — is free with no subscription. Paid is only the streaming engine: starting, pausing, and stopping migration or CDC runs.
Self-hosted. Desktop on macOS, Windows, or Linux; or Docker on a server. There is no SaaS tier; you control the deployment, the data, and the network boundary.
PostgreSQL and MySQL as both source and target, including managed services like AWS RDS, Azure Database, GCP Cloud SQL, Supabase, Neon, and DigitalOcean. File targets include CSV, JSON, JSONL, and Parquet; S3-compatible storage includes AWS S3, MinIO, DigitalOcean Spaces, Backblaze B2, and Wasabi. Snowflake as a target is in progress.
Yes — through the built-in MCP server. Your assistant reads live schemas, runs read-only SQL, and checks stream state directly from your workspace. Only SELECT statements pass the server-side filter, and nothing leaves your infrastructure. How it works →
Debezium provides log-based CDC but expects Kafka and orchestration around it. Airbyte and Fivetran are SaaS-style ELT platforms focused on connector breadth, not on operational database-to-database use. DBConvert Streams is self-hosted, focused on PostgreSQL and MySQL workflows, and bundles exploration, migration, and CDC into one workspace — no Kafka, no external orchestrator. See full comparisons →
The IDE is free. Streams run on a pay-as-you-go plan.