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
releaseRelease

Wren Core Python 0.7.4 fixes a release gap that stranded a newer semantic engine

A seven-line metadata change shows how multi-package release automation can leave Python users behind even after the underlying engine ships.

By The News Desk· Aug 13, 2026the quick take — two AI hosts go live when you do

The release is small; the failure it exposes is not

Wren AI published wren-core-py 0.7.4 on Aug. 11. Its release note contains one fix: declare which semantic-engine version the Python bindings build against. There is no new query feature or interface in this release. The practical change is that the Python binding can now carry the already-released wren-semantic-core 0.3.1 engine rather than leaving that engine inaccessible through the project’s Python release line. Wren Core Python 0.7.4 release Wren AI PR #2659

The underlying engine release appeared on Aug. 10. Wren describes wren-semantic-core 0.3.1 as fixing row-level access-control cycle detection so that its state remains local to each analyzer invocation. But the project’s own postmortem in pull request #2659 says the newest published Python wheel was still 0.7.3 and had been built against the pre-0.3.x engine. In other words, the Rust-side fix existed, while Python consumers had no release path to receive it. Wren Semantic Core 0.3.1 release Wren AI PR #2659

Why the automation missed it

Wren’s repository packages the semantic engine and its Python bindings separately. According to the merged pull request, the Python package depended on the engine only through a local path. The project’s release automation attributes changes to packages by file location, so fixes inside the engine directory triggered an engine release but not a Python-binding release. The automated engine-version bump was also categorized as a release chore, which the release tool skips when deciding whether to open another package release. Wren AI PR #2659

The repair was deliberately narrow. Commit 92cdf43 added an explicit 0.3.1 version alongside the existing local-path dependency and registered that field in Wren’s release configuration. The commit changed seven lines across two files—six additions and one deletion—and left local path resolution intact. Wren AI commit 92cdf43

That makes 0.7.4 less a feature release than a distribution correction. It also leaves follow-up work: the pull request says the manual metadata fix unblocks one release but does not fully automate future Python and WebAssembly binding releases after engine updates. For operators, the lesson is straightforward: in a polyglot semantic-layer stack, a tagged core library does not prove that every language binding contains it. Release provenance belongs in the deployment checklist alongside API compatibility and model behavior. Wren AI PR #2659

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.