Cogent and OpenAI Launch Cyber Defense Factories for Enterprises

Cogent, a North America-based technology startup, has partnered with OpenAI to launch Cyber Defense Factories aimed at automating vulnerability remediation for enterprises. The initiative integrates OpenAI’s advanced models, including GPT-6.1 Sol and GPT-5.6, with enterprise-specific contexts to streamline the process of identifying and addressing security risks.

AI-Driven Vulnerability Management

The Cyber Defense Factories are designed to help enterprises manage vulnerabilities at machine speed, addressing the growing challenge of accelerated vulnerability discovery and exploitation. According to Cogent, their AI agents can autonomously assess exposures, execute permitted changes, and verify that risks have been eliminated. This approach has reportedly reduced the mean time to remediate vulnerabilities by 75% and increased the number of vulnerabilities triaged by a factor of five.

Cogent’s Context Engine plays a pivotal role by maintaining an enterprise knowledge graph that includes assets, vulnerabilities, identities, ownership, and dependencies. This shared understanding enables the AI agents to operate efficiently across various security tools and systems. Additionally, GPT-6 Astra is utilized for upfront planning and complex reasoning, allowing the system to reuse context and reduce latency, inference costs, and inconsistent results.

Policy-Driven Autonomy

To ensure secure and compliant operations, Cogent’s Agent Policies define the permitted changes and approval requirements for each enterprise. The TRACE policy-enforcement engine verifies tool calls against these policies before execution, while an Agent Gateway independently checks authorization. High-risk tasks are executed in isolated sandboxes, and enterprises can choose to delegate investigations, require approvals, or allow bounded autonomous remediation.

Strategic Vision

Geng Sng, Co-founder and CTO of Cogent, highlighted the transformative potential of OpenAI’s models in enterprise defense. “The ability of OpenAI models to sustain long-running tasks unlocks a new level of autonomy in enterprise defense: AI agents can now carry work from investigation through verified remediation,” Sng stated.

Cogent also conducts internal research through its Cyber and Daybreak platforms, exploring attack paths and enhancing defensive capabilities. The company’s overarching goal is to enable enterprises to run their own Cyber Defense Factories, where AI agents handle the full lifecycle of vulnerability management, from disclosure to verified fixes.