The AI-native data pipeline platform.

rsync.ai is an AI-native data pipeline platform. You describe what to sync in plain English; an LLM agent discovers the schemas, scans for likely personal data and drafts a plan. Nothing is created or started until you confirm it, and then the pipeline runs as a durable Temporal workflow. Once the data lands you can shape it with Models, query it in the SQL Explorer, chart it, put it on a dashboard, and schedule reports and alerts to Slack or email, all in the same product. Use it as a managed cloud at app.rsync.ai, or self-host it free for internal use: it's source-available under the rsync.ai Source-Available License at github.com/rsync-ai/rsync.ai.

The catalog has 18 connectors. PostgreSQL, MySQL, SQL Server and MongoDB stream changes with real-time CDC. BigQuery and Databricks are live warehouses; AWS S3, Google Cloud Storage and Azure Blob Storage cover object storage; Google Sheets and Shopify cover SaaS. Oracle, ClickHouse and Notion are in the catalog too, and Snowflake, Redshift, Stripe and GitHub are in preview. For an API that isn't in the catalog, the AI Tool Generator on rsync.ai Cloud builds a versioned connector from its REST, GraphQL or OpenAPI docs in minutes, with auth, schema discovery and pagination included.

Yes. A model turns a SQL query into a table that rebuilds on a schedule or when what it reads has been refreshed. A chart draws a model or saved query as a bar, horizontal bar, line, area, heatmap, scorecard or table, and a dashboard puts charts and text on one live page with up to four filters. A workflow runs a query, checks a condition, and sends a message, chart or dashboard to Slack or email, on a schedule or after a pipeline. Ask rsync.ai, or your own AI tool over the agent gateway, can draft a workflow or dashboard, and you review it before it is turned on.

Yes. rsync.ai runs an MCP server, the agent gateway. Claude Code, Cursor, Gemini CLI and other MCP clients connect with a scoped, expiring token and can read your pipelines, runs, schemas and models, and draft pipelines, workflows and dashboards for a person to review. It returns metadata, not row values, and every call is audited. On a self-hosted install the gateway ships switched off until you turn it on.

Yes, or use the managed cloud. rsync.ai Cloud is live at app.rsync.ai with nothing to deploy. To self-host, install the stack in your own VPC with Docker Compose in one command (curl -sSL https://raw.githubusercontent.com/rsync-ai/rsync.ai/main/install.sh | bash); a Kubernetes install with Helm is in preview. Self-hosting is free for your company's own internal use under the rsync.ai Source-Available License. Credentials are AES-256 encrypted at rest with a key you control, and Ollama is supported for fully local LLM inference. AI connector generation is available on rsync.ai Cloud only.

Yes, for databases. rsync.ai uses log-based change data capture (Debezium) for PostgreSQL, MySQL and SQL Server, plus change streams for MongoDB, so changes reach your destination continuously instead of on a polling schedule. SaaS sources such as Shopify sync on a schedule you choose, as often as every minute.

Before data moves, rsync.ai scans your columns for likely personal data (emails, phone numbers, IDs, addresses) and suggests a rule for each field. You choose mask, hash (SHA-256 or HMAC), drop or pass-through, and the rules apply on every run before data lands. The Shopify connector also leaves customer PII out of its default GraphQL queries.

Fivetran is a closed-source managed service that bills on Monthly Active Rows. Airbyte has a large connector catalog that you configure connector by connector, and its self-managed version now runs on Kubernetes (abctl or Helm). rsync.ai starts from a plain-English request: the agent discovers schemas, proposes PII rules and a schedule, and builds nothing until you confirm the plan. It bills on GB moved, not rows, runs as a managed cloud or self-hosted, and on rsync.ai Cloud it can generate a connector for an API that isn't in the catalog. Models, charts, dashboards and scheduled reports are built in.

Data teams that want Fivetran- or Airbyte-style pipelines they can also run on their own infrastructure, analysts and ops leads who would rather describe a sync than script one, AI engineers who want their AI tools to reach pipelines over MCP, and smaller companies that can't justify per-row ETL pricing. If you can describe a sync in a Slack message, you can run it on rsync.ai.