Share with your CISO
Israeli startups pulled in $1.518 billion across 29 rounds in July, with cybersecurity capturing the largest share and a strikingly consistent thesis running through it: the enterprise attack surface has fundamentally changed because AI agents now act autonomously inside corporate systems, and the tools built to stop malware aren’t equipped for that. Glow raised $180 million at a $1.2 billion valuation targeting endpoint prevention, Onyx raised $113 million on agent security four months out of stealth, and Neo raised $100 million protecting autonomous agents in enterprise environments.
What this means for your business
The pattern here isn’t a funding trend, it’s a diagnostic. When nine separate companies, all launched within roughly the same window, independently converge on the same problem framing, that’s a signal the incumbents have a gap. If your security stack was designed before your organization deployed AI agents that can access sensitive data, make decisions, and call external APIs, you’re already running on a threat model that predates the threat. The question isn’t whether to care about this category. It’s whether you’re exposed right now while the vendors are still early.
The crowding in this space deserves scrutiny rather than reassurance. Glow, Onyx, Neo, Mate Security, Bloom Security, Hush Security, Oak, and Act Security are all circling adjacent problems: agent identity, agent behavior monitoring, non-human access control, and autonomous threat response. That’s too many companies for one market to absorb at scale, which means consolidation is coming, probably fast, and whoever you pilot today may get acquired, pivoted, or defunded within 18 months. The recurring failure mode in security vendor selection is committing architecture budget to a category leader before the category itself has settled.
The vendor to watch most carefully here isn’t any individual startup. It’s your existing identity and access management provider. Non-human identities, meaning the machine accounts and agent credentials that AI systems use to act on your behalf, are the actual control plane these startups are all trying to own. If your current IAM vendor isn’t on record with a credible roadmap for machine identity governance by your next renewal cycle, that’s a sharper signal than any of July’s term sheets.
Concept deep-dive: Non-human identity
A non-human identity is the digital credential an AI agent, automated script, or software service uses to authenticate and act inside your systems, analogous to an employee badge but issued to software instead of a person. These identities now outnumber human users at most large enterprises, often by an order of magnitude, and they’re typically managed with far less rigor. When an AI agent can read your CRM, send emails, and call financial APIs, its credential is a high-value attack target with almost no behavioral baseline to detect misuse against.
Based on reporting from Israeli startups raised $1.5 billion in July as investors doubled down on enterprise, originally published 2026-08-02 03:13:00.
