A New Company Just Raised $100 Million to Put Guard Rails on AI Inside Big Businesses

Key points

• Neo, a cybersecurity startup, raised $100 million across seed and Series A funding rounds as of 2025.
• Backers include Andreessen Horowitz and Bessemer Venture Partners, two prominent Silicon Valley investment firms.
• The company's focus is controlling and securing AI software that large organisations deploy internally.
• Neo emerged from stealth, meaning it built its product in private before going public, only this week.

A new security company called Neo just came out of hiding with $100 million in the bank and one job to do: stop enterprise AI from becoming a liability before companies even notice it has.

Funding came across a seed round and a Series A (a later-stage investment), with money from Andreessen Horowitz and Bessemer Venture Partners among others, as first reported by SecurityWeek. Those aren't names that bet early on uncertain ideas, which tells you something about how seriously the investor class is taking the problem Neo is pitching. It's the biggest raise we've tracked in this space since Valarian's $50 million round on 14 July.

What problem is Neo actually solving?

Big companies are deploying AI tools fast, and governance, meaning the rules that decide what those tools can see, do, and say, is lagging badly.

In practice, this plays out in uncomfortable ways. An AI assistant gets connected to internal documents it shouldn't touch. A coding tool trained on company source code leaks patterns a competitor could use. An automated workflow makes decisions about customers without a human checking the output. None of this requires a criminal to cause harm. The failure mode is speed: companies adopt AI faster than their security and legal teams can keep up.

Neo is positioning itself in that gap. The pitch is a platform that gives organisations visibility into what their AI software is doing and lets them enforce policies around it: which cloud services it can call, what data it can read, who can approve its actions.

This is genuinely hard to do. AI applications don't behave like traditional software. They make probabilistic decisions, meaning they don't follow a fixed set of steps every time. Standard security tools, the kind built to watch databases and web servers, weren't designed for this.

The space is getting crowded. Several well-funded startups are making nearly identical claims, and the big cloud vendors, Amazon Web Services and Microsoft Azure, are building their own AI governance features. Neo will need to show it can do something those platforms cannot, or won't, do for customers. Our shadow AI story from 17 July found that executives are actually the likeliest people to ignore the policies Neo would enforce, which is a harder problem than any platform solves.

If this category struggles, the post-mortem will say the market was sold control before the product existed to deliver it.

Your IT team should have a ready answer to this: what data can your AI tools access, and who signed off on that list.