Snowflake MCP server: ask questions, get governed answers

Snowflake AI analytics: what you can ask

Typical questions from data and analytics teams. Each answer shows the definition, source and query behind it.

• What was net revenue by region last quarter, using our approved revenue definition?
• Which product lines had the largest margin change month over month?
• Who are the top 20 customers by revenue this year?
• How many new customers did we add each month over the past year?
• Compare order volume by channel this quarter versus last quarter.
• Which customers have declining usage over the last three months?

Worked example: a governed Snowflake answer

One question, and everything the answer shows so you can check it. The figures are invented sample data.

SELECT c.region,
       SUM(f.amount_usd - f.credit_usd) AS net_revenue
FROM finance.fct_invoice_lines f
JOIN finance.dim_customer c ON c.customer_id = f.customer_id
WHERE f.invoice_date >= DATE_TRUNC('quarter', DATEADD(quarter, -1, CURRENT_DATE))
  AND f.invoice_date <  DATE_TRUNC('quarter', CURRENT_DATE)
  AND NOT c.is_test_account
GROUP BY c.region
ORDER BY net_revenue DESC;

Common pitfalls when asking Snowflake questions

Each one is settled once in an approved definition, so every person and every assistant gets the same answer.

• Several tables called revenue

  Raw, staging and mart layers often hold similar tables with different logic. The definition pins the table and columns, so every answer reads the same one.

• Time zones in timestamps

  TIMESTAMP_NTZ, TIMESTAMP_LTZ and TIMESTAMP_TZ treat time zones differently, so "yesterday" can shift by hours. The definition states the reporting time zone.

• Semi-structured columns

  VARIANT columns need explicit paths and casts. The definition records them once instead of every question guessing the JSON shape.

How Nexatron connects to Snowflake

Whose permissions

Questions run through the connection's Snowflake account, so its grants are the outer limit. Inside that, Nexatron roles and the row policies your administrator defines decide what each person can see.

Read and write

Read-only. Nexatron reads Snowflake data to answer questions and never writes to Snowflake.

Setup

An administrator connects a service user with key-pair (recommended) or password authentication. Steps: choose the source → check prerequisites → enter connection details → verify → ask. No data migration and no pipeline to build: data is queried where it lives.

Snowflake MCP connector for Claude, ChatGPT and Copilot

Nexatron's hosted MCP server is a governed Snowflake MCP connector: Claude, ChatGPT and Microsoft Copilot ask questions of your Snowflake data through it. Each person signs in as themselves, so the assistant sees only what they may see.

Connect Snowflake to Claude, ChatGPT or Copilot

1. Connect Snowflake to your Nexatron workspace.
2. In Claude, add Nexatron from Claude's connector directory. In ChatGPT or Copilot, add a custom MCP connector with the server URL https://api.nexatron.io/api/v1/mcp.
3. Sign in with your own Nexatron identity (OAuth 2.1), then ask your Snowflake questions.

Snowflake connector FAQ

Can I ask Snowflake questions without writing queries?

Yes. The connector is available now. Ask in Teams, Slack, the Nexatron web app, or an AI assistant connected through Nexatron's MCP server.

Does Nexatron respect Snowflake permissions?

Yes. The Snowflake account's grants are the outer limit, and your row policies narrow it for each person.

Can Nexatron change data in Snowflake?

No. Nexatron only reads Snowflake data.

What do I need to connect Snowflake?

A service user and an administrator to connect it.

Is there a Snowflake MCP server for Claude, ChatGPT or Copilot?

Yes: Nexatron's hosted MCP server, once Snowflake is connected to your workspace. It is Nexatron's server, not one published by Snowflake, and the assistant gets the same governed answer as the Nexatron app.

How is Nexatron different from Snowflake's built-in AI features?

Snowflake's AI features run inside Snowflake for people who work there. Nexatron puts governed answers in Teams, Slack, the web app and connected assistants, uses your approved definitions, and can combine Snowflake tables with other connected systems in one answer when federation is on (the default for new workspaces).

Can business users get Snowflake answers without a dashboard or SQL?

Yes. They ask the question; Nexatron writes the SQL from the approved definitions and shows it with the answer, so an analyst can review it. Existing Snowflake dashboards keep working alongside it.

Which Snowflake role does Nexatron use?

The connection uses one service user and role that your administrator chooses, and that role's grants are the outer limit. Within it, Nexatron roles and row policies decide what each person can see.