How I built agent-based security reviews on Databricks
Summary
An agent-based security review layer on Databricks automates predictable evaluations while routing ambiguous or high-risk cases to human reviewers. Built using Unity Catalog, Lakeflow Jobs, Databricks Apps, and hosted foundation models, this governed architecture improves review cycle times and consistency without removing human authority.
Summary generated by brickster.ai. For the full article, follow the source link above.
More from Databricks Blog
How Concurrence governs clinical AI at a trillion-token scale with Unity Gateway
Concurrence scaled its clinical AI agent platform to an annualized 1.2 trillion tokens by consolidating its data and AI architecture on Databricks. The system pairs an immutable patient world model built on Delta tables and Lakebase with Unity Catalog and Unity Gateway to enforce compliance-gated model routing and continuous agent evaluation.
The Genie One MCP is now Generally Available
The Genie One MCP server is now generally available, bringing trusted insights across structured and unstructured data into any AI agent workflow. Grounded in Genie Ontology, it provides agents with governed access to unified business context across tables, documents, and tools to reduce conflicting answers and support queries, visualizations, and citations.
Genie One MCP: Give any AI Agent the Right Business Context
Genie One MCP brings governed business context to any MCP-compatible AI assistant, enabling external agents to reliably interpret metrics and respect user access rules. Powered by Genie Ontology's approved definitions, data relationships, and source authority, it delivers reliable, permission-aware answers directly inside the tools teams already use.
Unity Catalog Pages: a governed home for your business knowledge in Genie Ontology
Unity Catalog Pages provide a governed home within Genie Ontology to define and organize key business terms, entities, and acronyms right next to the data they describe. This semantic layer grounds agents like Genie One in authoritative organizational context and can be rapidly populated from sources like Confluence, Slack, and Google Docs using Genie Code.
RADAR: Catch gray failures with anomaly detection
Gray failures often hide from standard dashboards, quietly eroding revenue and customers until RADAR's four-stage pattern—reliability metrics, anomaly detection, alerting, and root-cause analysis—surfaces them within minutes at over 90% precision. Built and proven on Databricks' own infrastructure, the same metric-agnostic system can be assembled with native components and an AI-agent scaffold to monitor billing, conversion, or model performance.
Enabling secure, productive work on personal devices
Databricks lays out a layered mobile security strategy that protects corporate data on employees' personal devices without giving up their privacy or trust. It spans device management, identity and access, zero trust, and application management, and the post walks through applying it to the Genie mobile app with Engineering as "customer zero."