Check Point presents Agentic Network Security Orchestration Platform, its agentic cyber orchestration platform

Check Point has unveiled its latest innovation: the Agentic Network Security Orchestration Platform. This groundbreaking platform is designed to automate network security operations within enterprise settings without the need for ongoing human intervention. In conjunction with this announcement, Check Point has also acquired the team and intellectual property of the Israeli firm Deepchecks, known for its expertise in evaluating and monitoring multi-agent systems.

Traditionally, enterprise firewall management has operated on a stable framework characterized by distinct vendor consoles, layered static rules accumulated over the years, and human teams manually reviewing policies in response to change requests. This model has served well in the centralized networks of the 2000s but has begun to falter since the mid-2010s due to the rise of hybrid cloud infrastructures, mergers and acquisitions that fragment security perimeters, the explosion of connected devices, and now the proliferation of artificial intelligence agents within systems.

The competitive landscape is marked by a shift among cybersecurity vendors towards agent-based architectures, a trend that gained momentum over the past year with contributions from companies such as Palo Alto Networks, Fortinet, Cisco, and CrowdStrike in adjacent segments like incident response and vulnerability management. Check Point is now stepping into the specific domain of network security, where slow review cycles have become bottlenecks for major transformation projects. The company reports that processing a simple change request can take anywhere from two to four weeks, with segmentation projects often stalled for years.

A Platform Built on a Proprietary Network Knowledge Graph

The Agentic platform is built around a proprietary Network Knowledge Graph that continuously models the client’s actual environment, including topology, traffic flows, asset dependencies, and configuration data. This enables agents to reason based on real-time network conditions rather than static training data. A layer of semantic intelligence interprets the syntax of existing firewall policies along with the business intentions behind them, including rules that may have been drafted years ago.

Four key capabilities structure Check Point’s offering. The Intent-to-Policy feature translates business requirements expressed in natural language into risk-validated firewall rules across multi-vendor environments. The Zero Trust and Policy Hardening capabilities analyze live traffic to identify orphaned accesses and overly permissive configurations, applying validated recommendations without disrupting network connectivity. Autonomous troubleshooting diagnoses incidents through multi-step reasoning over topology, policy history, and logs. Continuous compliance associates each rule and change with DORA, PCI-DSS, and NIST frameworks in real-time.

Acquisition of Deepchecks for Agent Evaluation Layer

Check Point has finalized an agreement to acquire Deepchecks, a platform specializing in evaluation, observability, testing, and monitoring of multi-agent systems. The acquired team consists of experts in large language models from Israel’s Talpiot program. The terms of the deal have not been disclosed.

This acquisition aims to accelerate the deployment timeline for agent orchestration within Check Point’s perimeter by incorporating the internal layer for measurement and continuous improvement. This strategy has become standard in the cybersecurity sector over the past eighteen months, as agent observability and traceability of decision-making have shifted from being optional features to non-negotiable requirements for production deployments, particularly under regulations like DORA and NIS2. The capabilities gained from Deepchecks specifically cover this area.

Distinction Between Co-Pilot, Autonomous Agent, and Operational Safeguards

The promise of autonomous execution requires careful attention from security leaders. The system described is neither a co-pilot requiring continuous oversight (where every action is validated by an operator) nor a fully autonomous agent (where every action is executed without approval). Check Point outlines a hybrid architecture, employing multi-step agents under predefined safeguards, enabling teams to define business intent and approve high-impact changes, while routine execution occurs automatically with full traceability. This distinction is not merely a matter of terminology; it fundamentally shapes the mapping of responsibility among system operators, the agent layer provider, and business stakeholders.

A portion of the offering is available immediately, including Policy Auditor for preventing policy drift, Policy Insights for Zero Trust hardening, and AI Assist for streamlining daily administrative tasks. The Playblocks agents can be accessed through an Early Availability program, with a broader client preview set to roll out in the second half of 2026—offering more agents, additional capabilities, and extended multi-vendor support.

Insights for European IT and Security Leaders

For security teams across France and Europe, the focus is shifting from technical decisions to governance considerations. The technical capability of an agent to harden a firewall policy or diagnose an incident within minutes—a task that used to take hours—is now well-documented, both by Check Point and its competitors. The operational questions revolve around the scope of autonomy granted to agents, the quality of traceability required by European regulatory obligations (DORA for financial entities, NIS2 for essential and important entities, AI Act for high-risk systems), and the internal skills needed to supervise a layer of agents in production.

Check Point Software Technologies, based in Israel and publicly traded on NASDAQ, frames the management of configuration data and network knowledge graphs outside of European jurisdiction. While mapping rules to DORA, PCI-DSS, and NIST frameworks provides technical controls to aid compliance, it does not equate to legal compliance in the European context. The increasing integration of agent-based layers in cybersecurity tools from major Anglo-Saxon and Israeli vendors raises parallel debates regarding cybersecurity sovereignty—discussions that have yet to yield effective answers among pure-play European firms.

Scroll to Top