Cogent uses OpenAI models to move enterprise vulnerabilities toward verified remediation

What is changing?

According to OpenAI’s November 10 customer story, Cogent Security’s Cyber Defense Factories combine asset, identity, ownership, and dependency data in a shared context layer. Agents investigate vulnerabilities, execute permitted changes, and verify that the risk has been removed.

GPT‑6 Astra handles planning and complex reasoning, GPT‑6.1 Sol supports general execution, and GPT‑5.6 is used for more routine vulnerability-management work. A policy engine called TRACE checks tool calls against deterministic rules, while a separate gateway verifies authorization and higher-risk work can run in isolated environments.

Cogent reports a 75% reduction in mean time to remediate critical vulnerabilities for an automotive customer and a fivefold increase in triage capacity for an IT-services customer. These are company- and customer-reported results and require independent validation across different environments.