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
analysisEnterprise agents

Workday’s Sana Coach turns employee signals into advice—but it is still an early-access system

The new agent can assemble reviews, structure spoken feedback and turn Peakon sentiment into suggested actions. Buyers still need evidence that those recommendations are fair, useful and controllable.

Employee signals assembled into coaching advice.
AI-generated illustration
By The News Desk· Oct 14, 2026the quick take — two AI hosts go live when you do

Workday has put Sana Coach into early access, extending its agent push from hiring and cross-application assistance into performance management. The product combines goals, check-ins, feedback, development activity, skills, organizational structure and sentiment in a conversational coaching interface. That is a consequential design choice: the agent is not merely drafting prose from a blank prompt; it is interpreting a broad record about workers and teams. Workday

What shipped

Workday says the early-access release focuses on three areas. For performance reviews, Sana Coach drafts, summarizes and structures inputs drawn from goals, check-ins, feedback and development activity. For feedback, it gathers comments from multiple reviewers, identifies themes and turns them into coaching prompts and development actions. Through Workday Peakon Employee Voice, it also converts sentiment signals into leadership insights intended to prompt intervention rather than remain in a survey dashboard. Workday

The agent accepts spoken as well as written feedback. Workday describes a turn-by-turn voice conversation—not simple dictation—that asks follow-up questions and structures the result into feedback ready to share. The voice and avatar experience was built with ElevenLabs. Workday

The boundary buyers should test

The release moves an enterprise agent closer to evaluative judgments about people. Workday says the system continuously gathers signals, surfaces themes and recommends next steps. It also says the guidance is grounded in governed Workday data, including performance history, skills and sentiment. But the launch post does not publish an evaluation of recommendation quality, subgroup consistency, reviewer-bias amplification or how often managers reject the agent’s suggestions. Workday

That missing evidence matters more here than fluent drafting. A summary can be edited; a recurring theme or suggested intervention can shape how a manager interprets an employee. Early customers should therefore test provenance for every generated theme, require a human decision before feedback or actions are committed, and measure disagreement and override rates across teams. Those are deployment recommendations, not controls Workday documents in the launch post.

Availability also limits the claim. Sana Coach is in early access. Workday describes leadership coaching as the next capability, while automated career paths and skill-gap analysis are planned for a future release. The current announcement demonstrates a direction and a working interaction model, not broad availability or validated workforce outcomes. Workday

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