Query your Snowflake data with AI — step by step. We'll show you both the manual approach and the faster way with Contextary.
This absolutely works. It's how most teams start. Here's what you need to do.
Get your Snowflake account URL, username, password, warehouse name, database, and schema. If your org uses SSO or key-pair auth, you'll need to sort that out too. Check with your Snowflake admin if you don't have direct access.
Open Claude and start writing a system prompt that explains your database. List every table, every column, what they mean, and how they join together. For a typical analytics warehouse, this can be hundreds of lines. You'll need to describe things like "the orders table has account_id which joins to accounts.id, not accounts.account_id."
Export your schema definitions and paste them into Claude's context window. For large Snowflake deployments with hundreds of tables, you'll likely hit context limits. You'll need to decide which tables to include and which to leave out — and hope the ones you left out aren't needed.
Add all the tribal knowledge that isn't in the schema: "status='complete' means the order is both paid AND shipped", "amounts are stored in cents, divide by 100", "exclude accounts where is_test = true", "fiscal Q1 starts in February". If you miss something, Claude will silently get it wrong.
When Claude generates a query, you'll need to manually copy it, open the Snowflake console (or your SQL client of choice), paste the query, select the right warehouse and role, and execute it. Check the results for obvious errors before trusting them.
Take the query results from Snowflake and paste them back into Claude so it can analyze, summarize, or visualize the data. For large result sets, you'll need to truncate or aggregate first. This copy-paste loop is where you'll spend most of your time.
Added a new table? Renamed a column? Changed a business rule? You'll need to update your system prompt manually. And remember to do it — stale prompts produce stale answers, and Claude won't know the difference.
Every person on your team will need their own copy of this system prompt. They'll inevitably customize it, use different definitions, and get different answers to the same questions. When the VP asks "what was revenue last quarter?" and gets three different numbers from three people, nobody knows which one is right.
But you'll spend more time managing prompts and copying SQL than actually getting answers. And every person on your team has to set this up independently.
Same result, fraction of the effort. Here's the entire setup.
Enter your Snowflake account URL, warehouse, database, schema, and credentials. That's it — 60 seconds. Contextary handles the connection, discovers your tables, and maps your schema automatically.
Every table, column, data type, and relationship is cataloged automatically. No copy-pasting DDL, no context window limits, no deciding what to include and what to leave out.
Document metric definitions, gotchas, and business rules in Contextary's handbook. AI even suggests most of it — you just confirm or tweak. Define "revenue" once, and every AI tool uses that exact definition.
Add one line to your Claude configuration to connect it to Contextary via MCP. Claude now has access to your full schema, business context, and the ability to run queries directly.
Ask Claude anything about your Snowflake data. Contextary provides the context, Claude writes accurate SQL, Contextary tells your warehouse what query to run, and you get the answer — all in one conversation. No copy-pasting, no switching tabs.
When someone adds a gotcha or updates a metric definition, every AI tool sees it immediately. No prompt maintenance, no context window limits, no re-explaining.
| Without Contextary | With Contextary | |
|---|---|---|
| Setup time | Hours of prompt engineering | 60 seconds |
| Schema changes | Manual prompt updates | Auto-discovered |
| Business rules | Copy-pasted into prompts | Documented once, shared everywhere |
| Team consistency | Everyone writes their own prompt | One source of truth |
| Metric definitions | Explained each time | Defined with exact formulas |
| Query execution | Copy-paste between tools | Contextary runs it for you |
| Maintenance | Ongoing prompt management | Set it and forget it |
Connect Snowflake to Claude in 60 seconds. Free to get started.