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
newsUNDERREPORTED

Alation made its regulated lineage interface AI-free on purpose

Critical Lineage unifies automated, manual and placeholder evidence now; AI comes later, after customers clear their own approval gates.

Before-and-after lineage graph, AI added only after governance approval.
AI-generated illustration
By The News Desk· Aug 31, 2026the quick take — two AI hosts, this story only

Alation has made Critical Lineage generally available for Alation Cloud Service customers, but the most instructive product choice is what version one leaves out: AI.

The capability sits inside Alation Critical Data Manager and builds a single, version-controlled graph from a critical report back to its source. It combines automated technical lineage with steward-entered manual links and explicit placeholder nodes for gaps such as spreadsheets, legacy applications and mappings that cannot yet be verified. Alation says the product reached general availability on August 13 and uses existing Alation Consumption Units.

That sounds like conventional governance plumbing until the interface decision comes into view. Alation says the first steward workflow is chat-based but deliberately AI-free. Its stated reason is deployment friction: regulated customers can begin using the product without triggering internal AI approval processes that typically add three or more months. The architecture is intended to add AI capabilities to the same interface later.

Build the evidence layer before the agent

Critical Lineage treats incomplete knowledge as something to govern rather than hide. A steward can register a spreadsheet as a catalog object, add an application that has no connector, or insert a placeholder when the exact mapping is unavailable. The placeholder records a known break in the chain until someone replaces it with a verified mapping.

At certification, the system captures a snapshot. Teams can then compare the current graph with that point-in-time record as systems migrate, applications are retired or report structures change. Provenance distinguishes automated edges from manual and placeholder lineage rather than presenting every link as equally certain.

For analytics-agent teams, that separation matters. An agent should not receive a polished lineage diagram that conceals unresolved hops. It needs the graph, the provenance of each edge and the open gaps as distinct inputs. Alation’s implementation makes that evidence usable before an organization is ready to let AI modify or interpret it.

The rollout lesson

The practical pattern is broader than this release: do not make AI approval a prerequisite for cleaning up the evidence that AI will eventually consume.

A regulated team can first establish report ownership, register offline artifacts, document known gaps and freeze certified snapshots. Those controls are useful for audits even if no agent ever touches them. When an AI layer is later approved, it inherits a governed record instead of becoming the mechanism that creates the record.

There is an important limit. A placeholder is evidence that a gap is known; it is not evidence that the missing mapping is correct. Teams still need an owner, a resolution path and version history. The value is that uncertainty becomes inspectable rather than disappearing between a hand-drawn business diagram and an automated technical scan.

Alation positions Critical Lineage for regulatory filings and critical reports, including environments governed by banking, insurance and public-company reporting requirements. The sharper practitioner takeaway is architectural: make lineage complete enough to inspect before making it intelligent enough to act.

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.