A raises $37M to attack its customers' systems nonstop with AI before real hackers can

Autonomous offensive security: an AI-powered attacker that runs the full cyber offensive lifecycle against its customers' systems, finds real attack paths, then helps fix them before a real attacker arrives.

That the old model — a $30,000 simulated attack every six to twelve months — is dead, and budgets shift to AI agents that continuously break in and fix first.

$37M from Lightspeed Venture Partners, Cyberstarts, and angels including Wiz CEO Assaf Rappaport and Cyera CEO Yotam Segev.

Yossi Torati spent six years at Israeli incident response firm Sygnia responding to breaches across roughly 40 countries. His founding moment came when he saw attackers begin to use AI and concluded most of what the defense industry had built was about to become obsolete. Cofounders Omer Gull and Yuval Itzchakov carry cyber backgrounds from Check Point, Hunters and IDF Unit 8200.

A emerged from stealth on June 8, 2026 with $37M. In one early proof-of-concept it found 1.2 million sensitive customer records, including Social Security numbers, that had sat exposed for seven years, undetected by the customer's own tooling. Lightspeed partner Guru Chahal, who led the conviction, argues manual pentesting is 'effectively dead' because discovery-to-attack windows have collapsed to minutes. The continuous threat exposure management market A chases was estimated at $2.7B in 2025, projected to $7B by 2033, with funded rival XBOW having raised $120M at a $1B-plus valuation in March.

It inverts the security product format — from periodic reports to a continuous autonomous attacker — rather than adding AI to an existing category.

The trigger is dated and specific: Anthropic's April 8 Mythos release surfacing thousands of zero-days across every major OS and browser.

The money and validation are concrete: $37M from Lightspeed, Cyberstarts and the CEOs of Wiz and Cyera, reported as a Fortune exclusive.

The proof point is a found problem, not a demo: 1.2 million records including SSNs, exposed seven years, missed by the customer's own tooling.

When the attackers' tooling jumps a generation, the defenders' delivery format — a report every six months — stops selling. Sell defense at the same cadence as the threat.

As of June 8, 2026, A had just emerged from stealth with $37 million and enterprise customers in finance, healthcare, critical infrastructure and technology, with the count undisclosed. The capital is earmarked for scaling go-to-market and building out the product roadmap. Its market — continuous threat exposure management — was estimated at $2.7 billion in 2025, and better-funded competitor XBOW had raised $120 million at a $1 billion-plus valuation in March, so the category is contested but young.