| Category | AI-native analytics engine with an integrated data stack | Warehouse-native AI — natural-language-to-SQL over Snowflake semantic views |
|---|
| Who it is built for | AI product and analytics teams, plus anyone who works with data — spreadsheet users through ML engineers | Data teams already standardized on Snowflake |
|---|
| Primary interface | The agent you already use, over MCP — plus Workspaces, notebooks, reports and data apps for the people who want a UI. All of them consume the same model rather than being the place it lives | Ask in English inside Snowflake; semantic views authored in SQL and YAML |
|---|
| Modeling language | Malloy — a modeling and query language with imports, inheritance, and public and private members | Semantic views, declared in SQL DDL and YAML, native to Snowflake |
|---|
| What is open | Open core. Malloy is open source, and we build and maintain Malloy Publisher, the open-source server for Malloy models. Your model is code in your repository, and Credible is in the Apache Ossie ecosystem for semantic interchange | Vendor-owned, and Snowflake started the initiative now incubating as Apache Ossie |
|---|
| How the model gets built | A coding agent with open-source MCP tools and agent skills, capturing context from where it already lives | Autopilot generates a semantic view from your tables, hand-tuned afterwards |
|---|
| Governance | Governance as code. Access rules are annotations in the model itself — versioned, reviewed and composable like any other code | Snowflake’s own role-based access control, inherited |
|---|
| Materialization and caching | One annotation on the source, in the same file as the logic it materializes — no derived-table block, no rollup definitions, no refresh triggers, no orchestration run to schedule. Optional per source: query your own warehouse directly, or hold a source hot in Credible’s in-memory serving layer, which is how a team on Postgres gets fast serving without buying a warehouse to get it | Whatever Snowflake gives you — result cache, dynamic tables, materialized views. Not a Cortex concern |
|---|
| How an agent finds the right data | Search. Typed targets — source, dimension, measure, view, even a dimensional value — are matched against an index of the model and come back ranked, so an agent asks for what it needs instead of picking a model and touring it | Cortex Search retrieves dimension values by hybrid keyword and vector search, re-ranked. Which semantic view to use is configured, not searched |
|---|
| Where the model can be used | One model served to every surface — agents over MCP, plus APIs, an SDK, embedded dashboards, notebooks and HTML data apps. Results carry an interactive UI resource (MCP Apps, the official extension), so a client that supports it renders a real chart or table instead of the model re-narrating rows | Snowflake APIs and Cortex Agents — inside Snowflake only |
|---|
| Scale and portability | Built for globally distributed, high-availability workloads. Warehouse-agnostic on open-source Malloy — BigQuery, Snowflake, Postgres, MySQL, Trino, Presto and DuckDB, which also reads Parquet straight out of object storage including Azure Data Lake — so the model travels | Scales with Snowflake, and runs only on data that is in Snowflake |
|---|
| What you pay for | Usage, not seats — unlimited users on every plan, so adding people never changes the bill. Metered per organization on tokens, bytes processed, bytes served and hot storage, starting free. Your own agent’s tokens are never billed, and a query that runs on your own warehouse is not metered for the scan — only for the result it hands back | Snowflake credits — a flat AI credit rate for Cortex on top of the platform credits the account already burns |
|---|