live wire
▸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 says Cortex Analyst apps keep working, but recommends Cortex Agents; semantic views and verified queries carry overSnowflake Documentation▸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 says Cortex Analyst apps keep working, but recommends Cortex Agents; semantic views and verified queries carry overSnowflake Documentation
nl2sql.ai
analysisSAP CONNECT

SAP is pairing Joule Work’s rollout with an agent registry and a mastered-data layer

The launch connects natural-language access, agent building and cross-system governance—but SAP’s announcement leaves runtime enforcement details to deployment design.

Layered enterprise agent stack vs plain assistant, with governed data and agent registry beneath natural-language access.
Side by side: what changed
By The News Desk· Oct 16, 2026the quick take — two AI hosts go live when you do

SAP says Joule Work will roll out to customers later in October, placing natural-language access to business data and agent-driven workflows in the same desktop experience. The more consequential part of the announcement is the surrounding control plane: SAP AI Agent Hub and SAP Reltio are now generally available, while Joule Studio is described as “soon” to reach GA.

Three layers, not one assistant

SAP presents Joule Work as the user-facing orchestration layer. It can answer business questions using context from SAP applications, SAP Business Data Cloud and non-SAP sources, then coordinate Joule Agents across business functions. Reinhausen, an early adopter, says it sees value in making S/4HANA data accessible through natural language; that is a customer statement in SAP’s announcement, not an independently measured accuracy result.

The second layer is construction. Joule Work’s Develop capability opens Joule Studio, where users can describe agents in natural language or use pro-code tools. SAP says agents deployed from Studio inherit SAP security and governance. The company also says Studio can connect third-party systems through open standards and more than 11,800 pre-built business accelerators.

The third layer is governance and data identity. SAP AI Agent Hub is now GA and is positioned to govern agents, language models and MCP servers across SAP and non-SAP environments. SAP Reltio is also GA in Business Data Cloud, providing mastered views of customers, suppliers, products and employees across systems.

What practitioners should verify

The architecture is directionally useful because it separates the conversational surface from agent registration and trusted entity data. A natural-language interface cannot resolve duplicate customer records by itself, and an agent builder cannot substitute for a registry that inventories external MCP servers and models.

But the launch post does not spell out the runtime policy mechanics teams will need to audit: how permissions propagate into every downstream tool call, whether MCP capabilities are restricted beyond registration, how approval gates are expressed for write actions, and which logs join a user request to the final business-system mutation. “Inherits SAP security and governance” is a product claim; buyers should ask for the concrete authorization path and audit record before treating it as an execution guarantee.

That makes the practical takeaway narrower than “one platform governs every agent.” SAP has assembled the pieces for a layered enterprise-agent stack: mastered business entities, contextual question answering, agent development and a cross-environment registry. The deployment work is to prove those layers preserve user identity and policy all the way through execution.

sources

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.

    more from the desk