Northstar Security has emerged from stealth with $7 million in seed funding to build an agentic security platform designed to connect fragmented enterprise data and expose combinations of weaknesses that conventional, issue-by-issue tools can miss. Wing Venture Capital led the round, according to the company’s launch announcement.
The San Francisco-based startup, led by CEO Amer Deeba and CTO Ed Rowe, says its platform creates a real-time “Security Context Graph” from existing security and IT systems. The graph maps relationships among identities, assets, applications, vulnerabilities, configurations and business systems, rather than treating each alert or exposure as an isolated record. Northstar says the platform can ingest information from more than 70 tools, although the company has not disclosed independent performance testing or customer results.
That connected view is intended to identify what Northstar calls “toxic combinations”: situations in which individually manageable conditions create a more serious path to a sensitive system. An unmanaged device, a vulnerability and an overprivileged identity may each generate a different alert, for example, but their relationship could indicate a practical route to business-critical data. The risk model is a product capability, not a newly disclosed vulnerability, and Northstar has not identified a CVE, public proof of concept or active exploitation tied to its announcement.
Northstar also says its AI agents can query the graph, investigate relationships and initiate remediation workflows such as deploying controls, notifying users or restricting access. The company describes human oversight as part of those workflows. That distinction matters for defenders evaluating autonomous security products: connecting data can accelerate triage, but automated changes to identity, endpoint or network controls can also create availability and access risks if asset ownership and business context are wrong.
For security teams, the immediate takeaway is to view the product as a context and orchestration layer rather than a replacement for vulnerability management. Organizations considering such systems should validate data coverage, permissions, audit logging, rollback controls and the boundaries placed around agent actions before allowing remediation to run automatically. Northstar says the new capital will support product development, hiring and go-to-market efforts; further details on adoption, pricing and independently verified detection outcomes were not provided in the company’s announcement.
