Generated interfaces threaten BI and dashboard incumbents

The core idea, a certified metrics and permissions layer that AI agents query over MCP, is already sold by Cube, AtScale, dbt and Strategy. Snowflake Semantic Views also offer it natively, and the OSI standard is making definitions portable, so a standalone metrics layer for Snowflake or BigQuery customers would be entering a crowded market. The remaining openings are narrower. The first is checking and certifying the output of generated dashboards: verifying that each chart in a prompt-built view actually came from a certified metric, that its filters respect permissions, and that it carries a certified or uncertified label. One comparison notes that current vendors do not track a metric's lineage or ownership past the agent that uses it. The second is companies that mix Microsoft Fabric or Power BI with Snowflake, since Power BI does not support Snowflake Semantic Views and Microsoft is outside OSI. The third is BigQuery customers without Looker, because the searches did not surface a Looker MCP offering. Each of these could be absorbed as a feature by an existing vendor, so a new entrant would need to move fast or attach to a specific AI lab's UI generator.

• Cube (Cube Core and Cube Cloud)

  Open-source semantic layer that defines metrics, joins and access rules once and serves them over SQL, REST, GraphQL, DAX/MDX and an MCP server for AI agents, with row-level multi-tenant security. Per Cube's own comparison, paid tiers start around $40 per developer per month.

• AtScale

  Universal semantic layer that serves one metric definition to Power BI, Tableau, Excel, Looker and AI agents over MCP. Its MCP server was listed on the Databricks MCP Marketplace in December 2025, and a third-party source says its engine runs inside Snowflake Semantic Views in private preview.

• dbt Semantic Layer (MetricFlow) and dbt MCP

  dbt Labs' metrics layer with an official MCP server for agents to query governed metrics. This entry comes from prior knowledge, not the searches; community demos that expose dbt artifacts to MCP agents also exist.

• Snowflake Semantic Views and Snowflake-managed MCP server

  Metrics, dimensions and relationships stored as native Snowflake objects. Standard SQL querying reached GA in March 2026, a managed MCP server serves them to agents, and Looker, Sigma, ThoughtSpot and dbt integrate with them.

• Open Semantic Interchange (OSI)

  Finalized, Apache-licensed YAML standard for portable semantic models, backed by Snowflake, Salesforce, dbt Labs, Databricks and others. Microsoft has not joined.

• Strategy Mosaic

  Positioned as governed metrics for business users, data engineers and AI agents across every warehouse and BI tool. Its performance claims are self-reported.

• Atlan context layer

  Governance and catalog vendor pitching a context layer that governs the whole estate around semantic-layer metrics for AI agents.

Checked 2026-10-09