Ask rsync.ai + Explorer

Ask your data
in plain English.

Ask a question the way you'd say it out loud. Ask rsync.ai writes the SQL; you read it, then run it in the Explorer under your own role, with personal columns masked in previews. Save it, chart it, or export it — in the same product that moved the data.

TL;DR

Ask rsync.ai turns a plain-English question into SQL grounded in your real schema, and shows it marked Not run. You open it in the Explorer, a built-in SQL editor, and run it under your own role: previews are capped and masked, writes are role-gated and audited, and results export as CSV, TSV or JSON. It lives in the same product that built your pipeline, so you can query a dataset the moment it syncs.

Natural-language-to-SQL grounded in your real schema
SQL shown as Not run until you run it, under your role
Masked previews; writes role-gated and audited
Save, chart or export every result

Start freeRead the Explorer docs

Question in, reviewed SQL out

Plain English goes in; grounded SQL comes out — and you decide when it runs.

Ask in plain English

Open Ask rsync.ai with ⌘/ (Ctrl+/ on Windows and Linux) and type a question like "Orders by status last month on the shop database". On a connection or pipeline page, the question is about that page.

It reads your schema, not your data

Ask rsync.ai works from the names, types and relationships in your connections. By default the model never sees your rows — sample rows are off unless your workspace turns them on.

Review SQL marked “Not run”

The query comes back as SQL marked Not run, with an Open in Explorer button. Nothing executes until you have read it.

Run it in the Explorer

Run it under your own role. Previews are capped at 500 rows with personal columns masked, and you can save the query, chart it, turn it into a model, or export it as CSV, TSV or JSON.

Built for people who ask questions faster than they write SQL

The plain-English starting point gets you moving; the Explorer keeps you in control.

Natural language to SQL

Ask a question in plain English and get SQL back. You don't need to memorize the schema to explore it, and the same panel can also explain a failed run or draft a pipeline.

Schema browser and query tabs

The panel on the left lists the tables and columns of the connection you're working on. Foreign keys the database declares show as relationships; guessed ones carry an inferred tag so you can check the join first. Up to 10 query tabs stay open at once.

Your role decides what runs

A SELECT needs any member role; writes need an admin; DROP and TRUNCATE need an owner with a typed confirmation. The server checks, not the editor, and every write is audited.

Save, chart, export

Keep a query under a name, turn it into a chart or a model, export a result as CSV, TSV or JSON, or share it to Slack — without another pipeline.

The query surface and the pipeline are the same product

A standalone text-to-SQL tool assumes the data already landed somewhere — you still needed a pipeline, a warehouse, and an export step to get there. rsync.ai builds the pipeline and gives you Ask rsync.ai and the Explorer over the result, sharing one schema and one product. The moment a sync completes, you can ask it a question. That's the whole loop: generate the connector, move the data, and ask questions of it in plain English — without leaving the app. Then turn the answer into a chart on a dashboard, send it on a schedule, or ask the same questions from your editor through the MCP server.

Ask rsync.ai and the Explorer — frequently asked

What is the rsync.ai Explorer?

It's a SQL editor on top of your connections. You pick a connection, browse its tables and columns, write a query and read the result. Previews are capped at 500 rows with personal columns masked. From there you can save the query, turn it into a chart or a model, and export the result as CSV, TSV or JSON. It runs in the same product that moved your data, so you explore what you just synced without a separate BI tool.

How does natural-language-to-SQL work?

You ask in Ask rsync.ai, the assistant panel in the product. It works from your connection's schema, picks the relevant tables and columns, and writes the SQL. The query is shown marked Not run, with an Open in Explorer button. Nothing executes until you read it and run it in the Explorer, under your own role.

Do I have to trust the AI's SQL blindly?

No. The SQL is shown to you before anything runs, and you can edit it. When you run it, the server checks the statement against your role, one statement at a time, so a member without write access can't run a write. Personal columns are masked in the preview and every write is recorded in the audit log.

Does the AI see my data?

Not by default. The model is a third-party service, so it gets metadata only: your question, names and types of connections and schemas, pipeline and run states, and error text with literal values scrubbed out. It never gets query results, row counts or credentials. Sample rows are an opt-in that is off by default and needs both a platform setting and the workspace owner to switch it on.

What can I query with the Explorer?

The connections in your workspace — sources and destinations. Pick one, browse its schema and run SQL. On MongoDB you browse documents and run read-only find and aggregate queries instead of SQL.

How is this different from a standalone text-to-SQL tool?

Most text-to-SQL tools are a layer you bolt onto a warehouse you already loaded with some other pipeline. Ask rsync.ai and the Explorer live in the same product that built the pipeline and moved the data, so the moment a sync lands you can ask questions of it, with no export step and no second vendor. The same product also holds the models, charts, dashboards and workflows you build from the answer.

Can I run the Explorer on my own infrastructure?

Yes. The Explorer is part of the self-hosted stack, which is source-available under the rsync.ai Source-Available License and installs into your own VPC. Ask rsync.ai ships switched off in a self-hosted install; you turn it on with one environment variable and give it a Claude, ChatGPT or Gemini key.

Your next answer is a sentence, not a query.

Start free on rsync.ai Cloud, connect a source, and ask your data a question in plain English.

Start free Read the Ask rsync.ai docs

Live on rsync.ai Cloud today. Self-hosting is available too; Ask rsync.ai ships switched off there until you add a model key.