Cogent Security, a San Francisco company that builds AI agents to find owners for software vulnerabilities and fix them, announced a $42m Series A on 18 February 2026.
Bain Capital Ventures led the round. Existing investor Greylock Partners and Definition took part, and founders and executives from OpenAI, Abnormal Security and Datadog invested personally. The round brings Cogent's total funding to $53m, about six months after the company's official launch in July 2025. No valuation was disclosed.
Fortune, which published an exclusive on the round, said Greylock partner Saam Motamedi had led Cogent's earlier $11m seed round.
“Security teams are drowning in coordination work: chasing down system owners, writing tickets, proving fixes happened.”Vineet Edupuganti, CEO and co-founder, Cogent Security
The problem Cogent targets
Cogent's announcement said more than 48,000 new software vulnerabilities (CVEs) were published in 2025, 162% more than five years earlier. It argued that finding them is no longer the hard part, since scanners handle discovery. The slow, manual work comes afterwards: working out who owns each affected system, judging real-world risk, coordinating fixes across teams and proving the work was done.
Cogent's agents investigate vulnerabilities, determine system ownership, assess risk in the customer's own environment and produce remediation steps for engineers, then track the fix until it is verified. The company says customers cut the exposure window for critical risks by an average of 97%.
How it works
Fortune reported that Cogent sits on top of existing security tools rather than replacing them. It connects to scanners, to internal asset lists such as those in ServiceNow, and to cloud and endpoint security data, then pushes action through ticketing and patching systems.
Many customers begin with the agents handling investigation, prioritisation and routing while humans keep the final remediation step, Fortune said, and some later grant more autonomy in safer development environments. Cogent says every agent action can be tracked and replayed and runs within approval rules set by the customer. Fortune also reported that the product uses models from Anthropic and OpenAI to help write the code that fixes issues.
Team and customers
The founders are Vineet Edupuganti, chief executive, and Geng Sng, chief technology officer, who both came from Abnormal Security, and Thanos Baskous, who led infrastructure at Coinbase, where he ran large-scale vulnerability remediation. The company also lists researchers from Google DeepMind on its team.
Cogent said it has worked with dozens of Fortune 1000 companies since launching. Fortune reported a target of a tenfold increase in customers this year. The one customer quoted in the announcement was Justin Yoshimura, chief executive of CSC Generation, who said Cogent is central to the commerce consolidator's cybersecurity strategy given the pace of its acquisitions.
What the money is for
In the announcement Cogent said it would accelerate product development, expand enterprise deployments and keep building agentic security systems. Fortune added that the company plans to move beyond vulnerability management into other security operations and IT automation work, and to quadruple its go-to-market team.
Enrique Salem, a Bain Capital Ventures partner and former chief executive of Symantec, said the volume of vulnerabilities has outpaced what human-driven coordination can handle, and that Cogent's software lets teams do five times more with the same resources. He told Fortune that Bain had known Edupuganti since before he founded the company.
Company website: cogent.com