Data pipelines,
in plain English.
Describe what you want to move. rsync.ai discovers the schema, flags PII, asks for your approval, then runs the pipeline — with real-time CDC for Postgres and MySQL and AI-generated connectors for any REST or GraphQL API.
rsync.aiSound familiar? You're not alone.
One sentence.
A running pipeline.
Watch the agent turn a plain-English request into a running Shopify→Postgres pipeline — then query the synced data in natural-language SQL. All in under 20 seconds.
Three steps.
No SQL, no YAML, no DAGs.
If you can write a Slack message, you can run a production data pipeline on rsync.ai — backed by Temporal workflows, Debezium CDC, and OpenTelemetry tracing under the hood.
Describe the pipeline in plain English
Type what you want. The LLM agent parses the source, destination, cadence, and any constraints (mask PII, exclude tables, JSONB nested objects). No forms, no DAG editors.
Review the plan, approve the gates
rsync.ai discovers your schema, proposes destination tables, and runs a PII scan. Each step pauses on a human-in-the-loop gate — connection, tables, PII rules — until you approve.
Pipeline runs on Temporal, you watch
Pipeline runs as a Temporal workflow with checkpointed cursors and automatic retries. OpenTelemetry traces let you replay any past run — no engineer required to debug.
What's inside
rsync.ai
Six surfaces that work together: chat-driven pipeline creation, data exploration, an AI connector builder, schema discovery, live monitoring, and an MCP server generator.
Describe it. Approve it. Done.
Type a request in plain English. The LLM agent parses source, destination, cadence, and PII rules into a Temporal workflow with human-in-the-loop approval gates.
- LLM planner — heuristic + DAG strategies
- Temporal workflow under the hood (durable + resumable)
- Approval gates at every critical step
Everything a data team needs.
Source-available. Self-hosted.
rsync.ai gives analysts, ops leads, and data heads a full pipeline platform — natural-language setup, real CDC for relational sources, PII detection, and an AI connector builder for any API.
Anyone on your team can build pipelines
Your analyst, ops lead, or PM can create a data pipeline by chatting with rsync.ai. The LLM agent breaks the request into steps — no SQL, no YAML, no DAG editors.
SQL workbench with AI inside
CodeMirror 6 SQL editor with schema-aware autocomplete and NL→SQL. Click any table or column in the sidebar to insert it at cursor. Export to CSV, TSV, or JSON and send results to Metabase or Superset.
AI generates connectors for any REST or GraphQL API
Paste an API docs URL or OpenAPI spec. rsync.ai reads it, generates a versioned MCP connector with auth, schema discovery, and pagination, and ships a Docker image in minutes.
PII caught before a single row moves
rsync.ai scans every column for personal data — emails, phones, IDs, addresses — and asks for per-field rules. Choose mask, hash (SHA-256/HMAC), drop, or pass-through.
You approve, AI executes
Every critical step — connection, table selection, PII rules, schema changes — pauses for explicit approval via a human-in-the-loop gate. Nothing moves until you say yes.
Real-time CDC for Postgres and MySQL
Log-based change data capture via Debezium keeps Postgres and MySQL destinations in sync within seconds. SaaS sources sync on a schedule you describe in plain English.
See exactly what's happening, always
Live row counts, run history, error alerts, and OpenTelemetry traces (SigNoz integration included). Replay any past run from the event log to debug.
Self-hosted — your data stays yours
Run the full stack on your own infrastructure via Docker Compose. Credentials are AES-256 encrypted at rest, and Ollama is supported for fully local LLM inference.
Per-pipeline namespace isolation
Each pipeline writes into its own destination schema (e.g., shopify.orders, shopify_brand_b.orders) with collision detection and ownership gating — Fivetran/Airbyte-style multi-tenant safety.
Connect what you actually use
Built-in connectors for the databases, warehouses, and SaaS tools most teams need. Don't see yours? The AI Tool Generator builds a working connector from any REST or GraphQL docs URL.
Point the AI Tool Generator at any REST or GraphQL docs URL and it produces a versioned MCP connector — auth, schema discovery, cursor pagination, Dockerfile included.
Built like a modern
data platform.
Plain-English UX on top of a serious foundation — Debezium-based CDC, Temporal workflows, OpenTelemetry tracing, and source-available connectors you can read, fork, or rebuild.
Built for teams that care about security & control
Production-grade infrastructure, audit-friendly logging, source-available code — without per-row pricing or vendor lock-in.
Run rsync.ai
on your own stack.
Self-host in one Docker Compose command, or book a 30-minute live walkthrough. Plain-English pipelines, real CDC, source-available under the Elastic License 2.0.