Agentic Security: Real-Time Threat Detection with Lakewatch
Overview
| Experience | Virtual |
|---|---|
| Technologies | Lakewatch |
AI-driven attacks now operate at machine scale, with mean time to exploit collapsing from 23 days to just under 2 days. SIEMs as we know them today are going to disappear because this new agentic era you are dealing with unprecedented threats. That’s why Databricks announced Lakewatch, the Open Agentic SIEM. Instead of locking security data into proprietary systems with per-byte ingest licensing, Lakewatch lets teams store all their security telemetry in open formats, retain it for years, and run agentic AI across it. Security teams gain complete visibility across the enterprise and can deploy defensive security agents to automate threat detection and response at massive scale.
Session Speakers
Andrew Krioukov
/GM of Lakewatch
Databricks
Jason Pohl
/Head of BuiltOn Field Engineering
Databricks
Maria Zervou
/Chief AI Officer - EMEA
Databricks
Full Summary
Why National Australia Bank bet on a lakehouse‑native SIM for security analytics
National Australia Bank operates at a scale that strains traditional security tools: millions of customers, thousands of applications and pipelines, and deep regulatory obligations. The bank's leaders argue that legacy SIMs, built for predictable infrastructure logs and manual triage, cannot keep up with attackers who now move at machine speed. A lakehouse‑native SIM on the Databricks Platform lets security teams widen telemetry, analyze more data, and automate repetitive work without hitting cost or capacity ceilings.
FAQ
Lakewatch is a SIM built on the Databricks Platform. It standardizes common security data workflows and leverages the platform's scale and open formats so teams can ingest and analyze far more telemetry than legacy tools that impose technical and cost limits.
The core workflow mirrors big data analytics: ingest logs, normalize them with ETL, run real‑time queries to detect anomalies, and investigate alerts through iterative analysis. Positioning it this way brings the tools, skills, and automation that data and AI teams already use.
Agents automate repetitive, manual steps such as setting up ingestion, generating ETL and detection rules, and triaging routine alerts like impossible‑travel logins. Automation reduces queues and lets analysts focus on investigations that require human judgment.
Lineage and access control are essential. Because Lakewatch runs on the Databricks Platform, organizations can use Unity Catalog to govern who accesses overlapping datasets across security, risk, and analytics use cases without creating unmanaged copies.
Teams need more software and data expertise alongside traditional security skills. Moving to an agentic, data‑driven operation requires retraining and change management so analysts are comfortable with new workflows and responsibilities.