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
newsCOST CONTROL

Snowflake’s per-user AI quotas turn cost controls into access controls

The 10.29 release lets administrators cap individual AI-credit use and automatically block new requests across Cortex Agents, Snowflake CoWork, CoCo and AI Functions.

By The News Desk· Aug 19, 2026the quick take — two AI hosts go live when you do

Snowflake has made per-user quotas generally available, moving AI cost governance from account-wide monitoring toward individual enforcement. In the 10.29 release, administrators can set daily and monthly credit limits for each user and apply them to AI Functions, Cortex Agents, Snowflake CoWork and Snowflake CoCo, as well as track warehouse compute in a separate quota. For supported AI domains, Snowflake can automatically block access when a user reaches the limit and release the block when the quota cycle resets. Snowflake’s 10.29 release notes describe the GA rollout and its changes from preview.

What administrators can control

A quota can cover every user in an account or a subset selected through user tags, allowing teams or cost centers to share one policy while each user’s usage is evaluated independently. The same per-user limit applies to everyone inside a given quota; Snowflake does not support different limits for different users in that object or a pooled cap across the group. Daily cycles reset at UTC midnight, and monthly cycles follow the UTC calendar month. The quota documentation also says administrators can configure actual-spend or projected-spend thresholds for notifications and custom actions.

At GA, Snowflake added separate monthly or daily notification-threshold targeting and made end-user emails for blocks configurable. It also introduced an account-level QUOTA_ACCESS_BLOCK_HISTORY view and replaced the preview GET_ACTIVE_BLOCKS() method with GET_ACTIVE_BLOCKS_V2(). The preview-only usage and refresh-tier methods were removed because Snowflake now says quota evaluation runs within minutes for all accounts. Those API changes are listed in the release notes.

The operational catch

The controls are not request-time hard ceilings. Snowflake says enforcement is evaluated within minutes after spend is recorded, so a user can overshoot before a block lands; a single large AI-function request can overshoot by more than interactive use. Configuration changes can take roughly five to ten minutes to propagate. Blocking applies only to supported AI domains: warehouse credits can be monitored, but warehouse queries are not stopped by the built-in block action. Snowflake documents those limitations explicitly.

For teams rolling out data agents broadly, that distinction matters. The feature provides a native kill switch for runaway individual AI consumption, but administrators still need conservative limits, notification thresholds and monitoring rather than treating the quota as an instantaneous billing boundary.

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