Getting Started

Connect your tools to AI assistants through a single MCP endpoint.

DataToRAG gives your AI assistant access to Google Workspace (Gmail, Drive, Calendar, Docs, Sheets, Slides, Contacts, Tasks) and Atlassian (Jira, Confluence) through a single MCP server. Want to see it in action first? Try the playground, a built-in chat in your dashboard that runs real tools against your connected accounts, no client setup needed.

DataToRAG dashboard with Google Workspace and Atlassian connected, example prompts, and the built-in playground

Quick setup

Prefer video? This two-minute walkthrough covers the sign-in flow, account connection, and your first prompt against Workspace:

Setup is two connections, and both are required before your assistant can do anything:

  1. Connect your accounts. Sign in at datatorag.com/dashboard and connect your Google account (and Atlassian, if you use Jira or Confluence).
  2. Connect your AI client. This is the step most people miss: connecting an account in the dashboard does not connect your assistant. Your AI client (Claude, Cursor, ChatGPT) has to be pointed at the DataToRAG MCP endpoint separately. Pick your client below for exact steps:

Claude web

  1. 1. Open Customize → Connectors
  2. 2. Click "Add", then "Add custom connector"
  3. 3. Paste this URL:
    https://datatorag.com/mcp
  4. 4. Complete the sign-in when prompted

Once your client completes its sign-in, ask it something like "search my email for invoices from last month". The first successful tool call shows up in your dashboard within seconds.

Connect through Claude on the web

The full flow for claude.ai in the browser (Claude Desktop is identical):

1. Open the connector settings. In Claude, go to Customize, then Connectors, click Add, and choose "Add custom connector".

Claude web Connectors screen with the Add menu open and "Add custom connector" highlighted

2. Name it and paste the endpoint. Use any name you like and this URL, leaving the advanced OAuth fields empty:

https://datatorag.com/mcp

Add custom connector dialog with the name DataToRAG and the URL https://datatorag.com/mcp filled in, advanced OAuth fields left empty

3. Complete the sign-in, and tick the boxes. Click Add, and Claude redirects you to sign in with the same Google account you used on the DataToRAG dashboard.

Google's consent screen arrives with every box unchecked, and the Continue button works anyway. If you click straight through, the connection succeeds, DataToRAG shows as connected, and not one tool will work, because you granted permission to nothing. There is no error at any point. This is the single most common way a setup goes wrong.

Tick "Select all", or tick just the services you want, then click Continue.

Google's consent screen for DataToRAG MCP. Every permission checkbox is unchecked by default, including Select all at the top, and the Continue button is still clickable. Select all and Continue are both circled.

You can grant a subset and come back later, and connecting more services afterwards re-runs this same screen. If tools are missing in Claude, this is the first thing to check: open the DataToRAG dashboard, look at what your account actually granted, and reconnect if it is short.

4. Check the tools are available. Back in a chat, DataToRAG appears in Claude's connector list and its tools (like gmail_search and jira_search) are available to the conversation.

Try it in the playground

You don't need to connect an AI client to see what DataToRAG can do. The dashboard playground is a built-in chat that runs the same tools against your connected accounts, right in the browser. Ask it to search your email or list your open Jira issues and watch the tool calls it makes.

Anything that would change your data (sending an email, updating an issue) pauses for your explicit approval before it runs, so you can explore safely.

The playground is for trying things out. For day-to-day use, connect your own AI client with the steps above so your assistant carries your tools everywhere you work.

How it works

DataToRAG runs as a remote MCP server. When your AI assistant calls a tool like gmail_search, the request flows through our gateway to your connected Google account. Your data never touches our storage. Every operation is a pass-through to the Google API on your behalf.

One account per sign-in email

DataToRAG makes one account for each email you sign in with. Your plan, your usage and your connected accounts all belong to that one account.

You sign in to DataToRAG in more than one place: on the dashboard, and once in each AI app when you add the connector there. Google shows its account chooser every time. Pick the same email every time. If you pick a different Google login in one app, DataToRAG makes a second account for it, on the free plan, and that app's calls count there instead of on the account you set up. Nothing warns you, and nothing fails until the free limit is reached.

Using more than one Google account does not need more than one DataToRAG login. Sign in with one email, then connect your other Google accounts from the dashboard under that login. Several connected accounts side by side are part of every plan, and you choose which one is the default.

If an app is already on the wrong account, the fix is in the Usage docs, under "Calls counting on the wrong account".

Authentication

Each user connects their own Google account through OAuth. DataToRAG requests only the scopes needed for the tools you use. You can connect multiple accounts and set a default.

Available connectors

DataToRAG currently supports these services:

Google Workspace

  • Gmail: Search, read, send, reply, forward, draft, save attachments, and manage labels
  • Calendar: List, create, update, delete events, and check availability
  • Drive: Search files, read content, create folders, rename and copy files
  • Docs: Read, create, write, and batch update documents
  • Sheets: Read, create, update, append, and delete spreadsheets, and manage tabs
  • Slides: Read, create, batch update, and delete presentations
  • Contacts: Search, list, create, update, and delete contacts
  • Tasks: List task lists, list tasks, create, update, complete, and delete tasks

Atlassian

  • Jira: Search issues with JQL, create, update, transition, comment, and look up attachment details
  • Confluence: Search with CQL, read and edit pages, read and add comments, look up attachment details

Each connector is documented on its own page with available operations, required scopes, and example prompts.

Seeing what your assistant is doing

Every tool call shows up in your personal usage dashboard at datatorag.com/dashboard/usage: call volume, latency, error rates, and a per-tool breakdown. See the Usage docs for what gets tracked and how retention works.

Common questions

The FAQ covers the questions we hear most: setup order, what the gateway can and cannot do, the approval gate, and how it compares to Claude's native connectors.

Frequently asked questions

I connected my Google account but none of the tools work. What went wrong? #

Almost always the consent screen. Google presents it with every permission box unchecked and the Continue button works anyway, so clicking straight through completes the connection, shows DataToRAG as connected, and grants permission to nothing. There is no error at any point. Open the DataToRAG dashboard, look at what the account actually granted, and reconnect, ticking "Select all" or the services you want.

Is connecting my Google account in the dashboard enough to set up DataToRAG? #

No. Setup is two separate connections and both are required: connect your accounts in the DataToRAG dashboard, then point your AI client at the DataToRAG MCP endpoint. Connecting an account in the dashboard does not connect your assistant, and this is the step most people miss.

What URL do I point my AI client at? #

The DataToRAG MCP endpoint is https://datatorag.com/mcp. Add it in your client as a custom connector, leaving the advanced OAuth fields empty, and the client then redirects you to sign in with the same Google account you used on the DataToRAG dashboard.

Can I grant only some Google services and add the rest later? #

Yes. You can grant a subset on Google's consent screen and come back later, and connecting more services afterwards re-runs the same screen. If tools are missing in your client, check what your account granted on the DataToRAG dashboard and reconnect if it is short.

I pay for Pro, so why are my calls counting on the free plan? #

You most likely have two DataToRAG accounts. DataToRAG makes one account per sign-in email, and each AI app asks you to sign in to DataToRAG once, when you add the connector. Pick a different Google login in one app and that app is on a second account, usually a free one, and its calls count there. Nothing errors until the free limit. To fix it, remove the DataToRAG connector in that app, add it back, and sign in with the email that holds your plan. Do not disconnect the Google account under the other login to tidy up, because that can break it under the login you keep. The Usage docs have the steps for each app.

Can I try DataToRAG without connecting an AI client? #

Yes. The dashboard playground is a built-in chat that runs the same tools against your connected accounts in the browser, with no client setup. Anything that would change your data, such as sending an email or updating an issue, pauses for your explicit approval before it runs. The playground is for trying things out; connect your own client for day to day use.

Where can I see what my assistant actually did? #

Every tool call appears in your personal usage dashboard, with call volume, latency, error rates and a per-tool breakdown, and the first successful call shows up within seconds. The Usage docs cover what gets tracked and how retention works.

Try this on your own account

Connect a service once and your assistant can run every tool on this page against your real data. Free to start, no card.

Get started free

By continuing you agree to our Terms of Service and Privacy Policy.