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

IBM Planning Analytics removes the all-or-nothing choice for custom agents

Workspace 3.1.11 lets teams add their own agent while keeping IBM-managed analysis features active, splitting customization from model operations.

Customer agent and IBM-managed analytics working in parallel.
AI-generated illustration
By The News Desk· Sep 25, 2026the quick take — two AI hosts go live when you do

IBM has changed the deployment choice behind its Planning Analytics Agent. In Planning Analytics Workspace 3.1.11 and 2.1.24, customers can connect a custom agent while continuing to use IBM-managed AI for the product’s built-in analytical features.

That sounds like a small configuration change. It is a meaningful operating boundary.

From one AI mode to two cooperating layers

IBM says the Planning Analytics Agent previously ran in one of two modes: fully IBM-managed or fully customer-managed. A team that wanted to bring in a custom agent—IBM gives a watsonx Orchestrate agent as its example—also had to provide the large language model used by every Planning Analytics AI feature.

The new hybrid model separates those decisions. A customer can build an agent around its own workflows, terminology or external data sources and connect it to Planning Analytics Workspace. At the same time, IBM-managed AI can continue powering the native exploration reports, outlier detection, impact analysis and other built-in capabilities.

For platform teams, the practical gain is not merely “more agents.” It is the ability to customize the business workflow without taking on model provisioning for the entire planning product. The release does not eliminate integration work: teams still need to decide what the custom agent may access and where requests should be routed. But it removes the earlier coupling between adding a custom agent and operating the models behind IBM’s native features.

MCP widens the integration path

IBM recommends watsonx Orchestrate as the starting point for custom agents, but it also says the Planning Analytics Agent’s MCP tools can be used with agent platforms that support remote MCP servers and OAuth. That gives teams a standards-based route beyond IBM’s own agent builder while retaining an authenticated connection path.

The first availability is in Planning Analytics Workspace. IBM says support for the hybrid model in Planning Analytics for Excel is coming soon, so teams should not assume the two client surfaces have feature parity yet.

Reports also get a presentation upgrade

The same release improves the agent’s exploration analysis report, a time-stamped PDF generated from a data view. IBM says the report now surfaces a wider range of patterns, uses clearer narrative language, and adds a table of contents, page breaks and layout improvements.

Those changes improve distribution, not validation. A polished PDF can make agent output easier to review and share, but the release post does not publish an accuracy evaluation for the new narratives. Teams should treat the report as an analytical starting point and keep their own review process for material planning decisions.

The larger release signal is architectural: IBM is no longer forcing customers to choose between a customized agent and the vendor-managed intelligence already embedded in Planning Analytics. That is a more realistic enterprise model, where teams often want ownership of the workflow without owning every model underneath it.

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