Startup helping secure agentic AI deployments emerges from stealth with $9M raise

As more companies adopt agentic AI, a DC startup just raised $9 million to help them deploy these new systems without compromising security.

Helmet Security finds, manages and monitors what are called MCP servers and connections. That stands for Model Context Protocol, an open source framework developed by Anthropic. In enterprises, MCPs can help LLMs access other company software so AI tools can act more like employees — doing things like sending emails and synthesizing data, explained cofounder and CEO Fred Kneip.

But AI agents are new on the scene, and deployment can accidentally give them more permissions than intended, opening the window for external hackers to manipulate the tools to expose sensitive information or send nefarious messages. 

Kneip believes companies aren’t prioritizing these risks enough, and said tech and security leaders have been expressing the need for more guardrails. That’s where Helmet Security comes in. 

“One of the scariest things for a security practitioner is to not know what’s going on,” Kneip told Technical.ly. “So we’re helping light that up for them to see, and then helping them implement controls and rules to limit the risk that exists.”

$9 million, plus connections to customers

With this seed round, led by SYN Ventures and WhiteRabbit Ventures, Helmet is officially exiting stealth. The funds will be used for R&D, Kneip said, and to expand the team from 3 to 10 employees, with a focus on engineering and customer management roles.

Helmet Security has been able to stay lean because of using AI tools in building its software and other operations, he said, noting that cofounder and CTO Kaushik Shanadi was able to build the product in a couple of months. 

The startup is also using AI to fill other business functions: Kneip said he doesn’t feel the need to hire a CFO as of now because of its capabilities. 

The two VC firms, known for investing in early-stage cybersecurity companies, are providing Helmet with more than just capital, he said. They’ve helped the startup with AI and cybersecurity expertise, and have connected the firm with potential customers. 

“Everyone on their team was the practitioner as well as an investor,” per Kneip, “and so they’[ve provided] really good feedback in that sense.”

The startup currently has about 10 customers and is in the process of onboarding 30 more to join in the next couple of months, he said. Though he declined to name any, he described them as a mix of commercial firms and government agencies. 

Helping engineers feel more comfortable with smart AI deployment

Kneip acknowledged the market is saturated with tools aimed at cybersecurity in AI systems, but said most focus on one aspect of the process, like identifying where permissions are too lenient or adding guardrails for specific use cases. 

The difference with Helmet Security, he said, is that it’s a comprehensive tool. It offers continuous monitoring of agentic AI systems and will update rules and protocols on an as-needed basis as technologies evolve.

The tool can also be paired with existing infrastructure, per Kneip, making it simpler for companies to implement. 

“If I make this hard, no one’s going to adopt it. We have to make this easy,” Kneip said. “You have to make it easy for the developers, the people who are building these agents.”

His ultimate goal? To give these agents the least amount of privileges to still be able to do the job, therein giving engineers more flexibility to innovate and deploy new AI capabilities at their companies. 

“[Helmet Security] gives that developer, the conscientious developer, the ability to move faster,” Kneip said. “It really can empower [them] to bring those tools even faster.”