live wire
IBM makes watsonx Orchestrate AgentOps, custom LLM judging and Bedrock-agent discovery generally availableIBMSchemaGate 0.1.45 fixes broken Oracle ADB wallet connections and an OCI stack pinned 28 releases behindSchemaGatePDI’s Amazon Quick procurement agent grounds spend answers in vendor, category and contract contextAWS Business Intelligence BlogBigQuery’s ML.METRICS example returns 0.84 accuracy but 0.30 macro-F1 on the same 100-row classification queryGoogle Cloud BigQuery docsSchemaGate 0.1.44 auto-selects sentence embeddings, lifting bundled-schema retrieval from 90/98 to 93/98SchemaGateSchemaGate 0.1.43 adds read-only SQL execution with per-principal table checks—and documents unauthenticated client assertionsSchemaGateDatabox adds reusable AI Analyst Skills with personal/company scope, auto-matching and marketplace installsDataboxFabric previews an AI builder for data-agent instructions, source guidance and example queriesMicrosoft FabricDatabricks trains data-agent retriever to stop early or spend bounded extra search steps, reporting 5.8-second latencyDatabricksThoughtSpot adds SpotterCode coding agent to its Visual Embed PlaygroundThoughtSpotLongMemEval-S audit: 67–73% of restore-fixable 80k-budget errors came from evicted evidence under three policiesarXivSchemaGate 0.1.42 adds dimension-aware retrieval and fixes complex multi-table SQL promptsSchemaGateSnowflake agent toolsets can silently drop inherited tools when callers lack accessSnowflake DocumentationLooker’s VS Code extension reaches GA with MCP-assisted LookML generation, editing and validationGoogle Cloud Looker release docsIBM makes watsonx Orchestrate AgentOps, custom LLM judging and Bedrock-agent discovery generally availableIBMSchemaGate 0.1.45 fixes broken Oracle ADB wallet connections and an OCI stack pinned 28 releases behindSchemaGatePDI’s Amazon Quick procurement agent grounds spend answers in vendor, category and contract contextAWS Business Intelligence BlogBigQuery’s ML.METRICS example returns 0.84 accuracy but 0.30 macro-F1 on the same 100-row classification queryGoogle Cloud BigQuery docsSchemaGate 0.1.44 auto-selects sentence embeddings, lifting bundled-schema retrieval from 90/98 to 93/98SchemaGateSchemaGate 0.1.43 adds read-only SQL execution with per-principal table checks—and documents unauthenticated client assertionsSchemaGateDatabox adds reusable AI Analyst Skills with personal/company scope, auto-matching and marketplace installsDataboxFabric previews an AI builder for data-agent instructions, source guidance and example queriesMicrosoft FabricDatabricks trains data-agent retriever to stop early or spend bounded extra search steps, reporting 5.8-second latencyDatabricksThoughtSpot adds SpotterCode coding agent to its Visual Embed PlaygroundThoughtSpotLongMemEval-S audit: 67–73% of restore-fixable 80k-budget errors came from evicted evidence under three policiesarXivSchemaGate 0.1.42 adds dimension-aware retrieval and fixes complex multi-table SQL promptsSchemaGateSnowflake agent toolsets can silently drop inherited tools when callers lack accessSnowflake DocumentationLooker’s VS Code extension reaches GA with MCP-assisted LookML generation, editing and validationGoogle Cloud Looker release docs
nl2sql.ai
guideOPERATIONS

Audit Fabric connections before you put agents on top of them

Microsoft’s new recency fields separate configuration history from actual credential use—and expose three cleanup decisions that should not be automated blindly.

Two Fabric connections contrasted: configured vs actively used.
AI-generated illustration
By The News Desk· Sep 10, 2026the quick take — two AI hosts go live when you do

Microsoft Fabric administrators now have a better way to answer a deceptively hard question: is a shared connection actually in use, or is it merely still configured?

The new Connection Recency metadata exposes two different signals. LastBoundDateTime records when a connection was most recently linked to a Fabric item. LastCredentialUsedDateTime records when a workload most recently used the connection’s credentials. The fields are intended for governance and investigation rather than real-time monitoring.

That distinction matters before teams expand natural-language and agentic access to a Fabric estate. A connection can remain bound to an item that runs every day, or it can be newly linked and never execute. Treating “old” and “unused” as synonyms can either preserve forgotten credentials indefinitely or break a legitimate seasonal workflow.

Three reviews, not one delete rule

Microsoft’s companion notebook turns the metadata into three review queues: stale connections, duplicate connections, and ownership-continuity risks. The sample defaults to flagging credentials unused for 90 days, but explicitly warns operators to confirm annual, seasonal and incident-response workloads before changing anything.

For duplicates, the suggested comparison uses connection type, path, connectivity type and gateway ID, then uses recent credential activity to identify the likely preferred connection. For ownership, it flags connections whose only owner is an individual user and recommends an approved Microsoft Entra group as the more durable operating model.

The sample is deliberately conservative. Its executable audit cells are read-only; the optional ownership-change cell is commented out. Microsoft also notes that API visibility is caller-dependent: cloud connections are generally visible only to their owners, while gateway administrators can see connections on gateways they administer. A tenant-wide inventory therefore requires an identity and scope design, not just a notebook run.

The practical rollout order

Teams should use the new fields as evidence for a human-reviewed lifecycle process:

  1. Inventory the connections visible to the audit identity.
  2. Separate “last linked” from “last credentials used.”
  3. Ask owners to validate low-frequency workloads.
  4. Consolidate confirmed duplicates.
  5. Replace single-person ownership with approved groups where appropriate.
  6. Only then broaden self-service or agent access.

This is not an agent feature by itself. It is plumbing—and that is precisely why it matters. An analytics agent can inherit the fragility of every connection beneath it. Connection Recency gives operators a measurable way to reduce that hidden operational debt before adding another access surface.

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.