live wire
▸Snowflake’s Cortex Agent quickstart routes metrics and transcripts with prompt rules—but documents no authorization boundarySnowflake▸Oracle Select AI generates SQL/PGQ for property graphs—but errors if one AI profile mixes graphs with tables or viewsOracle▸Domo’s data team standardizes MCP builds with shared scaffolding; beta toolkits inherit user permissionsDomo▸Databricks proposes five outcome signals for agentic BI—and says dashboard counts and license usage miss decision qualityDatabricks▸One vibe-coded database app needed 56+ human test hours and three repair rounds despite AI Studio’s verification claimsMancas & Mancas / arXiv▸BigQuery AI.IF’s default cost mode trains a local model for most rows—and drops row-error thresholdsGoogle Cloud▸AWS’s multi-tenant data-agent pattern gives the broker Lambda no data permissions; Cognito roles carry Lake Formation scopeAWS Big Data Blog▸Amazon OpenSearch’s optimized engine adds nested agent traces, but AWS reserves it for log analytics—not vector searchAWS▸Two interpretation-based provenance semantics cannot express SQL bag difference, new theorem showsKolaitis / arXiv▸Databricks Feature Views reuse one definition for training and live scoring, but online lookup still returns latest available stateDatabricks▸Databricks Genie’s Slack and Teams previews default channel answers to public, even when viewers lack source-data accessDatabricks▸Snowflake’s two-tool GraphRAG agent beat a seven-tool graph setup 70% to 60% on 22 ontology questionsSnowflake Engineering▸Semantix opens an Apache-2.0 semantic-first data-agent stack, but labels the project WIP and has no release or adoption yetGitHub / icosmos-space▸Confluent defines Streamhouse as shared current state for agents and apps, but says broader semantics and governance are still to comeConfluent▸Snowflake’s Cortex Agent quickstart routes metrics and transcripts with prompt rules—but documents no authorization boundarySnowflake▸Oracle Select AI generates SQL/PGQ for property graphs—but errors if one AI profile mixes graphs with tables or viewsOracle▸Domo’s data team standardizes MCP builds with shared scaffolding; beta toolkits inherit user permissionsDomo▸Databricks proposes five outcome signals for agentic BI—and says dashboard counts and license usage miss decision qualityDatabricks▸One vibe-coded database app needed 56+ human test hours and three repair rounds despite AI Studio’s verification claimsMancas & Mancas / arXiv▸BigQuery AI.IF’s default cost mode trains a local model for most rows—and drops row-error thresholdsGoogle Cloud▸AWS’s multi-tenant data-agent pattern gives the broker Lambda no data permissions; Cognito roles carry Lake Formation scopeAWS Big Data Blog▸Amazon OpenSearch’s optimized engine adds nested agent traces, but AWS reserves it for log analytics—not vector searchAWS▸Two interpretation-based provenance semantics cannot express SQL bag difference, new theorem showsKolaitis / arXiv▸Databricks Feature Views reuse one definition for training and live scoring, but online lookup still returns latest available stateDatabricks▸Databricks Genie’s Slack and Teams previews default channel answers to public, even when viewers lack source-data accessDatabricks▸Snowflake’s two-tool GraphRAG agent beat a seven-tool graph setup 70% to 60% on 22 ontology questionsSnowflake Engineering▸Semantix opens an Apache-2.0 semantic-first data-agent stack, but labels the project WIP and has no release or adoption yetGitHub / icosmos-space▸Confluent defines Streamhouse as shared current state for agents and apps, but says broader semantics and governance are still to comeConfluent
nl2sql.ai
releaseProduct release

Oracle’s Deep NL2SQL treats business knowledge as a governed release artifact

Private Agent Factory 26.10 adds topic-level context, clarification, SQL review and pre-publish evaluation instead of letting user feedback rewrite shared knowledge automatically.

By The News Desk· Oct 26, 2026the quick take — two AI hosts go live when you do

Oracle has added Deep NL2SQL to AI Database Private Agent Factory 26.10, pairing SQL generation with a curated Business Knowledge Graph and a release workflow for changing that knowledge.

The design targets a familiar production failure: a database schema can tell a model where fields live without defining what the business means by an open invoice, completed order or top-selling product. Oracle’s new workflow groups tables, metrics, rules, documents, business definitions and example question-SQL pairs into topics. At query time, the agent retrieves relevant topics, plans the request and can ask a clarifying question when the request is ambiguous.

Oracle illustrates the intended scale with finance and supply-planning knowledge graphs spanning 350 and 386 tables or views, respectively. Those numbers describe examples shown by the vendor, not a published accuracy benchmark, but they make the target clear: this is aimed at large enterprise schemas rather than a small demo database.

Feedback does not publish itself

The most consequential part of the release is the maintenance path. Subject-matter experts can inspect natural-language questions beside reference SQL, review generated SQL, curate annotations and update semantic context in dedicated workbenches. Proposed knowledge changes can then be evaluated against representative questions and reference SQL before they are published.

That separation matters. A thumbs-down or corrected answer is evidence, not automatically trusted training data. Oracle explicitly positions feedback for expert review rather than allowing each user response to mutate the shared knowledge base. Teams can therefore treat a definition change more like a governed software release: propose it, run a regression set, inspect the SQL and publish only after review.

The announcement does not provide evaluation-set sizes, accuracy results or a public benchmark for Deep NL2SQL. Buyers should ask how regressions are scored, whether tests can be promoted across environments and how topic retrieval failures are distinguished from SQL-generation failures.

A broader private-agent runtime

Private Agent Factory 26.10 also adds a Python SDK and standardized REST APIs for embedding agents in applications and workflows. Oracle says supported authentication includes username/password and browser-assisted SSO, while operations retain the user’s existing Agent Factory permissions. Expanded agent-to-agent support adds JSON, Markdown and asynchronous tasks.

For operations teams, native observability exposes metrics, sessions, traces and spans, including latency, token use and errors. A reusable ingestion service prepares knowledge for multiple agent workflows, and OpenRouter support broadens model options for the Knowledge Agent.

Taken together, the release makes NL2SQL one component of a managed private-agent platform. Its strongest idea is not a new prompt technique; it is the decision to make business context inspectable, testable and publishable under human control.

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