Gurucul Previews Security for AI with Runtime Detection and Response, and Launches Advanced AI-Assisted Data Pipelines to Support a Cost-Optimized, Vendor-Neutral Data Fabric

New AI-assisted data onboarding and pipeline health monitoring in Data Optimizer, plus preview of our next-generation of Security for AI - all running on the same Unified Entity Intelligence behind Gurucul's Next-Gen SIEM and Insider Risk Management

LAS VEGAS, Aug. 4, 2026 /PRNewswire/ -- Gurucul today previewed new Security for AI capabilities at Black Hat USA 2026, extending its entity model to cover AI agents, copilots, models, tools, plugins and MCP servers as first-class entities alongside humans and machines. The company also announced two new capabilities in Data Optimizer, its native data pipeline management offering: AI-assisted parsing and pipeline health monitoring. Live demonstrations at Black Hat USA 2026, Booth #4912.

The new Security for AI capabilities run on Unified Entity Intelligence: Gurucul's continuously updated risk model for the identities, behavior, relationships, privileges and activity of every human, machine, application, service account, data asset and AI agent in the enterprise.

Enterprises are adopting AI agents faster than they can secure them, while security budgets strain under rising telemetry volumes. Gurucul is addressing both at once – extending its entity model to treat AI as a first-class entity, and cutting the cost and effort of getting security data in.

Gurucul has been recognized as a Leader in the Gartner® Magic Quadrant™ for SIEM.

Security for AI runtime detection and response

Enterprises are deploying AI agents, copilots, models, plugins and MCP servers faster than security teams can inventory them, let alone govern them. The tools already in the SOC were built to watch users, devices and applications – not autonomous agents that hold credentials, call tools and act without a human in the loop, and not AI-to-AI interactions where neither party is a person.

Gurucul is closing that gap with an expansion of Unified Entity Intelligence. The new capabilities discover and inventory the AI running across the enterprise, surface Shadow AI and usage that bypasses corporate governance and flag suspicious or malicious AI activity. Detections align to MITRE ATLAS as well as MITRE ATT&CK Framework, and confirmed activity can trigger automated response playbooks, revoking an agent's access or quarantining a workflow inside boundaries the customer sets. 

While there are many new specialized tools on the market, they focus on specific domains such as proxy, run-time or gateways. What makes this different is not just the breadth of signal in our coverage list; is that an AI agent is modeled as an entity with its own learned baseline, its own privileges and activities. An agent drifting beyond its mandate is caught by the same Behavioral AI and advanced detection logic that catches a compromised employee or an over-privileged service account. Static policy can only enforce what someone anticipated in advance. Behavioral AI reads what the agent actually does.

Gurucul fuses these detection modalities together in our Relational Graph, leveraging endpoint, proxy, network and CASB signals alongside ML-driven baselining to validate the risk and provide deep evidence and context as the risk is forming. This enables analysts and agents, with purpose-built playbooks, to work together and end the threat before it turns into a full-scale incident.

Data Optimizer enhancements enable more efficient, cost-optimized agentic data operations

As data growth surges exponentially, SOC teams are forced into untenable trade-offs between cost and visibility. In addition, onboarding this new telemetry is often laborious and slow, creating a new bottleneck for the SOC.

Gurucul's Data Optimizer was built to solve these challenges for customers by intelligently retaining only the data and attributes that carry security value with the option of retaining all data for a significantly lower cost. And it can run on the data lake the customer already owns, including Snowflake and Databricks, so there is no proprietary store to buy and no migration to fund enabling enterprises with an efficient cost-optimized vendor-neutral Agentic Data Fabric.

Two new capabilities have been added to create deep new efficiencies in onboarding and on-going maintenance. First, AI-assisted parsing reads previously unknown or unsupported log formats and builds out the parser to turn them into usable data with human supervision. This greatly speeds up the process and enables teams to onboard sources without waiting for vendor-built connectors or custom parser development. Second, pipeline monitoring detects parsing failures, schema drift, volume anomalies and ingestion latency before they degrade detection, so security coverage does not break silently and create new blind spots for the SOC.

These advancements reduce the operating cost, not just the storage bill. When parsing and pipeline health are handled by the system rather than by engineers, the total cost is reduced. In customer environments, Data Optimizer has reduced ingestion costs by 40% to 87%.

Why Entity Intelligence works: two kinds of systems, one layer

Gurucul's Entity Intelligence leverages a neuro-symbolic AI architecture. It pairs two things most products treat as alternatives.

Behavioral AI builds the understanding – self-learning baselines for every entity, peer-group analytics, relationship and access-path analysis, and adaptive risk models with explainable output. This is what catches what no rule describes: a novel technique, an insider whose intent is still forming, an agent whose behavior has drifted. In customer environments it delivers up to 70% fewer false positives than static-rule detection.

Deterministic Logic supplies the facts – static and correlation rules, YARA and signatures, MITRE ATT&CK and ATLAS technique matching, policy and privilege checks. Gurucul's SecOps platform ships with more than 5,000 out-of-the-box detection models and 98.3% MITRE ATT&CK coverage with out-of-the-box content. Precise and citable on anything already defined.

The Relational Graph binds them, so a rule hit and a behavioral anomaly on the same entity become one case with the evidence attached, instead of two tickets in two queues.

This enables grounded AI: generative and agentic AI do not re-derive the picture from raw alerts on every task. They read an understanding that has already been computed, with every contributing signal attached – which is why their conclusions can be audited against the evidence that produced them, and why they are affordable to run at scale.

"AI agents are rapidly transforming how enterprises operate, but organizations need deeper security intelligence to understand and manage the risks that come with autonomous systems," said Saryu Nayyar, Founder and CEO of Gurucul. "Gurucul's unified entity intelligence with behavioral AI, built on a vendor-neutral data fabric, gives security teams the context they need to understand activity, identify risk and respond faster. By extending these capabilities to AI agents, we are helping enterprises securely embrace AI while reducing operational complexity and strengthening their ability to prevent, detect and respond to threats in real time."

About Gurucul

Gurucul delivers AI-SOC platform and Insider Risk Management powered by Gurucul's Unified Entity Intelligence on a vendor-neutral agentic data fabric architecture. By continuously understanding the identities, behavior, relationships, privileges, and activity of every human, machine, data, and AI entity, Gurucul turns fragmented security activity into contextual, risk-prioritized, explainable intelligence. That intelligence detects the identity attacks and insider data movement that matter most, identifies the risks created by enterprise AI, and gives human analysts and SOC AI agents the context to investigate, decide, and act faster, on the stack and sovereign data lake the customer already owns. Gurucul technology is used by Global 1000 companies and government agencies globally. To learn more, visit https://gurucul.com/ and follow us on LinkedIn and Twitter.

Media Contact: Michael Grenetz, SVP Growth, [email protected]

Find Gurucul at Black Hat USA 2026: booth #4912

SOURCE Gurucul