Think inside your AI world.
How do I connect Snowflake to Claude?
Snowflake's MCP path runs through Cortex — Analyst for semantic querying, Search for retrieval — so Claude asks the warehouse in meaning, not just SQL.
Direct answer: connect Snowflake's MCP surface in Claude (Settings → Connectors, authorise; 'claude mcp add' in Claude Code). The path runs through Cortex Analyst and Cortex Search — semantic querying and retrieval — so questions can arrive as questions and be resolved against governed data models rather than hand-written SQL.
Semantic reads over governed data
How did activation move by segment, what does the revenue model say for the quarter, which documents match this search — resolved through Cortex against the models your data team governs. The semantic layer is the point: the question meets the model, not a raw table.
Two systems, one number — by whose rule?
Worked case: your team ratified a single-definition rule — 'activation' is defined once, in the warehouse model, and every surface cites that definition; set after a board meeting where two decks disagreed by eleven points. Snowflake hosts the definition; that every other tool defers to it is an organisational ruling no warehouse announces.
The read that carries the ruling
Through Unl the metric read arrives under the rule: this is the warehouse-model activation figure — the citable one, per your single-definition ruling; the product tool's variant is flagged as non-citable for decisions. The eleven-point meeting stops recurring, because the ruling now travels with the number.
Snowflake to Claude asks the warehouse in meaning; Snowflake through Unl adds whose meaning governs — your single-definition ruling arrives with every figure it covers.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. The reach lane is live: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
What is the Snowflake MCP surface built on?
Cortex Analyst (semantic querying against governed data models) and Cortex Search (retrieval) — questions resolved through the semantic layer rather than raw SQL access.
Does the same connection serve ChatGPT?
Yes — the MCP surface is client-neutral; other clients connect through their own settings with the same authorisation.
Why ratify a metric-authority rule if the warehouse already governs models?
The warehouse governs its own models; nothing in it says other tools must defer. That deference is an organisational decision — exactly the kind Unl holds, with the eleven-point story that made it necessary.
What if the warehouse's model and my ratified definition disagree?
That disagreement is the most useful thing the read can give you. One of the two is stale, and finding out at the moment a figure is used beats finding out in the meeting where it is quoted — the warehouse governs the data, your ruling governs which meaning your decisions are allowed to cite.
What this is
Think inside your AI world — you stay in command
Save the thoughts, decisions and targets worth keeping, each with its reasoning, carried into every AI session the moment they matter. A new unit of exchange between you and your AI: the Settled Why with standing that travels. Unprompted.
Your whole AI world. What you decided at the epicentre. It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector — quick to connect, in a couple of steps.
MCP native·Human settled·Model agnostic·Your data
Measured Context
Connect Unl to bring the right information into the moment.
Your sources, read against the criteria you set.
The full product, open. Free at launch.