Northstar Security Emerges With $7M for AI Risk Detection

What changed

Northstar Security has emerged from stealth with $7 million in seed funding led by Wing Venture Capital, with backing from MVP, Outliers and cybersecurity veterans Gerhard Eschelbeck, Renaud Deraison and Justin Somaini.

The startup is building AI agents around a “Security Context Graph” that pulls information from more than 70 enterprise security and IT products. It connects identities, assets, applications, vulnerabilities and business systems, then helps analysts investigate risks and, when authorised, take action such as restricting access or prompting users to install endpoint protection.

Why it matters

Northstar’s bet is that serious security exposure often hides between tools, not inside one of them. A privileged user, a critical payroll application and an unprotected laptop may each look like a manageable issue alone. Together, they become the kind of “toxic combination” the company wants its platform to find.

That makes the graph the centre of the product, more than the AI agents themselves. Northstar says it integrates with Okta, GitHub, Workday, Microsoft Defender and Entra, AWS and Google Cloud. Customers can also use the Model Context Protocol to let external AI agents access the graph and build custom workflows.

The practical promise is speed: an analyst could ask questions in plain language, connect facts spread across multiple systems and authorise a response from the same platform. The practical risk is equally clear. Giving AI agents permission to limit access or notify users turns security automation into an access-control problem of its own. The value will depend on how reliably Northstar distinguishes a dangerous intersection from an ordinary one.

CEO and co-founder Amer Deeba previously co-founded Normalyze and held senior roles at Qualys. CTO Ed Rowe previously led engineering at AppDynamics. Northstar plans to use the funding to grow its engineering team, develop the platform and accelerate go-to-market work. Customer numbers, detection accuracy, remediation results, valuation and deployment scale have not been reported.