live wire
▸Snowflake agent evals can pin judge versions; unpinned metrics may roll forward and shift scoresSnowflake▸ThoughtSpot makes Spotter 3 default for new instances; it requires download rights and always shares query data with an LLMThoughtSpot documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub▸Snowflake agent evals can pin judge versions; unpinned metrics may roll forward and shift scoresSnowflake▸ThoughtSpot makes Spotter 3 default for new instances; it requires download rights and always shares query data with an LLMThoughtSpot documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub
nl2sql.ai
newsEnterprise AI

Power Apps opens its AI-generated pages to SQL data — but the connector path is still preview

Microsoft’s generative-page builder is no longer confined to Dataverse, giving enterprise teams a direct route from natural-language UI requirements to apps backed by SQL and other Power Platform connectors.

Power Apps generative pages now connect to SQL through approved connector references.
AI-generated illustration
By The News Desk· Sep 24, 2026the quick take — two AI hosts go live when you do

Microsoft has put SQL and other external data sources within reach of its AI-generated Power Apps pages. In the September Power Platform update, the company said generative pages can now bind to existing Power Platform connector references rather than relying only on Dataverse tables. The connector support is in public preview. (Microsoft Power Platform)

What changed

A maker can describe a page in natural language, point the Power Apps agent at selected data, and have it generate the React-based interface and associated business logic. Microsoft’s documentation says the new connector route supports the Power Platform connector ecosystem, allowing data outside Dataverse to be used in a generative page; Microsoft’s launch post explicitly names SQL and SharePoint as examples. (Microsoft Learn, Microsoft Power Platform)

The setup is not a free-form database connection created by the model. An administrator or maker must first create and authenticate a connection, then create a connection reference in a solution. That reference becomes available under Add data → Connectors, where the maker selects the connector data the generated page may use. Microsoft says the same capability works with external AI code-generation tools, which guide the user through selecting a connector and add the binding to the deployable page. (Microsoft Learn, Microsoft Power Platform)

Why it matters for SQL teams

This is not a new text-to-SQL query engine. It moves natural-language generation one layer up the stack: the prompt produces an application page that can read from a governed SQL connection, rather than merely producing a query for an analyst to run. That makes the feature relevant to teams turning database-backed workflows into internal applications, while preserving an explicit connection-reference boundary between the generated experience and the underlying system. (Microsoft Learn)

The caveats are substantial. Microsoft labels connector support as preview and says preview features are not intended for production use and may have restricted functionality. In-browser authoring is currently limited to Power Platform environments in the United States, Great Britain, Australia, and Singapore. Microsoft also says makers remain responsible for validating generated code against their organization’s security, accessibility and compliance requirements. (Microsoft Learn)

For practitioners, the near-term test is therefore architectural rather than promotional: confirm which connector operations the generated page actually needs, review the generated code, and validate identity and data-access behavior before treating the preview as an application-delivery path. The important shift is that SQL-backed data can now participate directly in Microsoft’s agentic page-building workflow instead of requiring a Dataverse-first design. (Microsoft Learn)

Filed by The News Desk. Corrections: desk@nl2sql.ai · Our standards →

comments · 0

    Comments are moderated before they appear. Your email is used once to confirm it is you — never shown, never sold. Corrections and questions get an answer from the desk when we have one.