Setup Guide

Set up DataAssist-IO in about 10 minutes

Connect your data once, add one URL in your AI client, and your whole team can ask questions in plain English. No SQL, no engineers, nothing to install.

Set up DataAssist-IO in three steps A three-step setup strip. Step one: connect your data — files or live databases. Step two: add DataAssist-IO to Claude, ChatGPT, or any MCP client. Step three: ask questions in plain English and get read-only answers. A progress line fills from left to right, lighting each step with a checkmark before the loop gently resets. 1 Connect your data files & live databases 2 Add to Claude or ChatGPT or any MCP client 3 Ask in plain English read-only answers, in seconds DataAssist-IO
Typically under ten minutes. No engineers needed.
Where the ~10 minutes go “~” times are typical for a first setup — not a guarantee.
  1. 0:00Create your account
  2. ~2:00Connect a data source
  3. ~6:00Add the URL in Claude or ChatGPT
  4. ~9:30First answer
1

Create your organization

Go to app.dataassist.io/signup, enter your organization name and email, and verify your address with the code we send. That's the whole account setup.

💡
The first user becomes the org Admin. Admins invite more users, connect data, and manage settings. Every new org starts with 30 free query credits — no card required.
2

Connect your data

Open Data Sources in the sidebar and add a source:

  • 📁Upload your files — spreadsheets & scheduled file drops. They're stored in your organization's isolated workspace as columnar, analytics-ready tables in an open table format — encrypted, versioned on every refresh.
  • 🗄️Connect a live database — with a read-only login. Your database is never copied — queries run read-only, in place.

Then choose which tables to expose — only exposed tables are ever visible to your AI client.

ℹ️
Every source has a built-in setup guide inside the app — open any connector form and click “Setup guide” for the exact details (read-only login, network path, certificates). Browse everything we connect to on the integrations page.
3

Add DataAssist-IO to your AI

One URL is the entire integration. Pick your client:

Works in Claude on the web, Claude Desktop, and the Claude Code CLI.

  • 1Open Settings → Connectors → Add custom connector.
  • 2Paste the MCP server URL below and click Add.
  • 3Claude opens the DataAssist-IO sign-in — log in with your email + password (or a one-time code) and approve.
MCP server URLhttps://app.dataassist.io/mcp
ℹ️
No client ID or secret to copy. Claude registers itself automatically (Dynamic Client Registration). Access renews on its own — you won't be asked to sign in again for about 30 days.

Added through ChatGPT's connector / Apps settings.

  • 1Open Settings → Apps & Connectors and add a connector.
  • 2Paste the MCP server URL below; set auth to OAuth.
  • 3ChatGPT opens the DataAssist-IO sign-in — log in and click Allow.
MCP server URLhttps://app.dataassist.io/mcp
ℹ️
No API keys and no callback URLs to configure. ChatGPT registers itself automatically; its callback URLs are already allowlisted server-side.
Connecting DataAssist-IO inside your AI client in three steps A generic AI-client settings window titled Settings, Connectors. Step one: the URL https://app.dataassist.io/mcp is typed into an Add custom connector field and the Add button is pressed. Step two: a small OAuth window appears — the DataAssist-IO lockup, Sign in to DataAssist-IO, two input fields and an Approve button. Step three: back in the settings list a connector row appears showing DataAssist-IO with a green Connected chip, and three tool chips fade in beneath it: list_tables, query_data and get_sample_data. 1 one URL 2 secure OAuth sign-in 3 tools appear automatically AI client — Settings › Connectors Add custom connector https://app.dataassist.io/mcp Add Installed No connectors yet DataAssist-IO Connected ✓ list_tables query_data get_sample_data DataAssist-IO Approve
Add one URL in Claude or ChatGPT, sign in, and your data tools appear — that’s the whole setup.
4

Ask your first question

Start a new conversation. Your AI discovers your tables automatically — try one of these:

“What tables do I have?” “What were our top 5 products last month?” “Show me customers who haven't ordered in 90 days.” “What's our revenue trend this quarter?”
Ask a question in your AI assistant, get an answer from your live data A generic AI-assistant chat window that works with Claude, ChatGPT and any MCP client. The user asks “Which customers generated the most revenue?”. A small DataAssist-IO tool-activity chip shows a read-only query_data call running. The assistant replies with a five-row results table: Northwind Traders in Mumbai with 28,930; Sundar Textiles in Surat with 15,980; Bluebird Cafe in Pune with 11,190; Metro Supplies in Delhi with 8,725; Green Grocer Co in Bengaluru with 8,670. A footer notes the answer came from the live database and was logged for admins. Your AI assistant — works with Claude, ChatGPT & any MCP client you Which customers generated the most revenue? DataAssist-IO query_data · read-only assistant customer city revenue Northwind Traders Mumbai 28,930 Sundar Textiles Surat 15,980 Bluebird Cafe Pune 11,190 Metro Supplies Delhi 8,725 Green Grocer Co Bengaluru 8,670 answered from your live database · logged for your admins
A real question, answered from live data in seconds — one credit per question.
Then invite your team. Add users under Users (admins manage, viewers ask questions), and optionally create Teams to grant each team exactly the tables it should see. Every question lands in a structured audit trail with user attribution — who asked, what ran, when — on the Tool Calls page.
?

Troubleshooting

The database connection test fails
Check the host and port are reachable from the internet. If your database lives in a private network (VPC), turn on the encrypted, host-verified tunnel option in the connector form and provide your jump host and key. Encrypted connections (TLS) are on by default — paste your provider's CA certificate to verify the server's identity.
I didn't receive my verification email
Use “Resend code” on the verify page and check your spam folder. Sign-in stays blocked until the address is verified — that's intentional.
What happens when we run out of credits?
Queries pause with a clear message — nothing is deleted and your sources stay connected. An admin can buy a credit pack under Billing; the balance is shared by the whole organization and credits never expire.
Who on my team can see which tables?
Admins see every exposed table. A viewer who isn't on any team also sees all of the org's tables, while a viewer on one or more teams sees only the tables those teams were granted.
Will my team have to sign in again and again?
No — after the first OAuth sign-in, access renews automatically with refresh tokens (about 30 days at a time in Claude).
Can the AI change or delete our data?
No. Every statement is parsed and validated before execution — single SELECT-only statements, nothing else — so writes are rejected before they ever reach your data. The security overview has the full picture.