Horizon3 AI

Horizon3.ai starts from a blunt question that makes most security teams squirm: you spent millions on tools, so can you actually prove you're secure? Its answer is NodeZero, an autonomous pentester that safely runs real attack techniques against your live production environment, chains the weaknesses together the way a human intruder would, and shows you the exploitable paths that actually matter.

Founded in 2019 and run out of San Francisco, the company was built by a fusion of former U.S. Special Operations cyber operators, national security veterans, and frustrated industry practitioners. CEO Snehal Antani, previously CTO of JSOC and CTO at Splunk, likes to call the approach turning the map around: take the attacker's perspective, because production is where the hackers actually attack. That bet has scaled fast, past a quarter million autonomous tests run in production and thousands of customers including government agencies and Fortune 10 giants.

NodeZero isn't a vulnerability scanner, and Horizon3 is aggressive about the distinction. A scanner tells you a flaw might exist; NodeZero proves whether it's actually exploitable by safely exploiting it, then shows the blast radius on your business. It runs without agents and without disruption, pivoting through the network, harvesting credentials, and chaining misconfigurations into full attack paths in a real-time view.

The loop is the whole pitch: hack to prove exploitability, fix what genuinely endangers the business, verify the fix worked with a one-click re-test, then repeat continuously as the environment changes. The company frames it as moving from vulnerability management to closing proven attack paths, and it leans hard on evidence over opinion-based risk scores.

Security teams drown in noise. Scanners spit out paralyzing to-do lists ranked by theoretical severity, manual pentests are expensive and go stale the moment the environment changes, and nobody can say which of ten thousand findings would actually get them breached. Horizon3 built NodeZero to cut that Gordian knot by testing what's real, not what's hypothetical.

Its founders name the pains directly: ineffective tools, false positives, alert fatigue, blind spots, a checkbox security culture, the cybersecurity skills shortage, and the lead time and cost of hiring outside consultants. The company argues AI has handed attackers unprecedented scale and speed, and that the only credible defense is to fight AI with AI and validate continuously instead of once a quarter.

NodeZero was built for the people who own security posture and are tired of guessing: IT Ops and SecOps teams, in-house and consulting pentesters, and the MSSPs and MSPs who run security for others. Horizon3 sells across a wide size range, from small educational institutions to Global 100 enterprises and government agencies.

The named proof points skew high-stakes. The company says it's trusted by global governments, Fortune 10 giants, and major healthcare providers, and cites the NSA and four of the Fortune 10 among its users, exactly the environments where a test going wrong is not an option.

Everything Horizon3 sells is NodeZero, its self-directed penetration testing platform that you can set up and point at your first test in minutes, from a free Docker host or virtual appliance for internal tests, or straight from the Horizon3 cloud for external ones. It covers internal, external, cloud, and Kubernetes attack surfaces, plus Active Directory password auditing and phishing impact testing.

The platform is packaged into tiers that layer on capability: continuous scheduling, precision threat detection with auto-dropped honeytokens, emerging-threat Rapid Response backed by Horizon3's own attack team, and risk-based exposure management with high-value targeting and threat-actor intelligence at the top end. There's also a NodeZero MCP server, a nod to where the product is heading.

Horizon3 sells NodeZero as enterprise SaaS with unlimited-scope pentesting, so cost scales with capability and coverage rather than per-test fees. Pricing itself is quote-based through a demo, but the packaging tells the story: you climb from occasional testing to continuous validation to full exposure management as you move up the tiers.

The published lineup runs Flex (episodic autonomous pentests), Core (continuous testing with scheduling), Pro (Core plus Rapid Response and Tripwires), and Elite (Pro plus high-value targeting, advanced data pilfering, threat-actor intelligence, and vulnerability risk intelligence). A free trial account lets teams run their first internal pentest before talking to sales.

Horizon3's whole positioning is a wedge against the incumbents: static scanners that test only what you already know, breach-and-attack-simulation tools that model threats rather than prove them, and periodic manual pentests that are stale within days. Its counter is autonomous exploitation that learns and adapts with every test and delivers proof instead of opinion-based risk scores.

The strategic tell is in their recent writing, arguing the category isn't red teaming and that organizations buy answers, not activities. That's a company trying to redraw the map around exploitability-first validation, with a moat built on hundreds of thousands of production tests that, they claim, outpace the collective history of manual pentesting.

The headline stat is scale run safely: more than 257,000 autonomous tests executed in production with, the company says, zero downtime across all of them. That production-safe record is the credential it trades on with high-security customers, and it's backed by real accolades rather than self-description.

Horizon3 landed #1 in Security on the Inc. 5000, #3 overall on the Deloitte Technology Fast 500, a spot on Fast Company's Most Innovative Companies, and reported 102% ARR growth heading into 2026. One customer case study says NodeZero cut attack-path impacts from 251 to zero across 18 locations, the kind of before-and-after that's hard to argue with.

Reviewers consistently praise the thing Horizon3 built the company around: NodeZero behaves like an actual attacker, discovering, pivoting, and chaining issues into clear attack paths tied to business impact. People like that it needs minimal setup and no agents, runs fast across broad scope, and lets non-pentesters get useful results, with one-click verify to confirm a fix without rerunning everything.

The criticism is just as consistent. Reporting can read too generic or high-level instead of clean per-asset detail, Docker installation gets tricky inside locked-down enterprise networks, and results lean on how well the environment is scoped, so conservative IP ranges can hide paths. Some reviewers note occasional false positives or missed nuanced vulnerabilities, a fair reminder that autonomous still isn't infallible.

Horizon3 raised a $100M Series D led by New Enterprise Associates in June 2025, with SignalFire, Craft Ventures, 9Yards Capital, and Qualcomm Ventures joining, reportedly at a valuation north of $750M. Total funding sits around $178M to $183M depending on the source, from a backer list that also includes NightDragon, NewView Capital, and Bridgespan.

Then in January 2026 came a strategic investment from Prosperity7 Ventures, the diversified venturing fund of Aramco Ventures, a signal that Horizon3's global-security story is drawing sovereign-scale capital, not just Silicon Valley venture money.

Horizon3's thesis is that cyber warfare is heading toward algorithm-versus-algorithm at machine speed, with humans by exception, and it's building NodeZero to be the defender's side of that fight. The recent moves, Rapid Response, an MCP server, a European HQ, and Aramco-linked capital, all point the same direction: continuous, autonomous, exploitability-first validation as the new baseline.

The open question is whether autonomous testing fully wins the trust of the most cautious buyers, given the reporting-depth and false-positive gripes reviewers still raise. But with triple-digit ARR growth, a category it's actively trying to define, and marquee government and Fortune 10 logos, Horizon3 is arguing from a position of strength.

Horizon3 describes its people as learn-it-alls, a deliberate anti-know-it-all posture, drawn from technology startups, Fortune 500s, and the U.S. military. The founding DNA is unusual: special operations cyber operators and national security veterans sitting alongside startup engineers, which shows up in a mission-first, attacker-minded culture that prizes grit and ownership.

On the engineering side, the work is genuinely hard and they say so plainly, building an LLM-driven autonomous pentester under a strict no-false-positives, production-safe mandate where the hard problems are reliability engineering, not chasing benchmarks. It's remote-first with offices in San Francisco, Chicago, and a new Amsterdam EMEA headquarters, and the company took a rare public moment to mourn its late CFO, Holly Grey, on its own leadership page.

Horizon3 leans into pay transparency, publishing base ranges in its job posts to comply with state disclosure laws. Disclosed U.S. engineering bases run from roughly $102K for support engineering up to about $315K for senior leadership roles, with product and design landing in the $175K to $260K range and sales stretching wide from $80K to $450K once on-target commission is folded in.

Every full-time role is eligible for an equity package in the form of stock options, and the company states its comp philosophy is to pay fairly and consistently across roles, levels, and locations. Bands are also posted in EUR for its growing European roles.

The legal entity behind the product is Horizon3 AI, Inc., which discloses in its privacy policy that it does not directly collect or store payment card data and keeps customer data governed by contract terms separate from its general privacy policy. As a security company selling into government and regulated buyers, Horizon3 builds against the frameworks its customers live by.

Across its security-engineering roles the company names SOC 2, ISO 27001, GDPR, FedRAMP, NIST, CIS, and MITRE ATT&CK as the compliance and control frameworks its own program is held to, which doubles as a signal of the assurance posture enterprise buyers should expect.

The past year reads like a company pressing its advantage: a $100M Series D, a new Amsterdam headquarters to anchor EMEA, the launch of Rapid Response for AI-powered attacks, and NodeZero clearing awardable status for U.S. Department of War work in the Tradewinds marketplace. Fast Company named it to its 2026 Most Innovative Companies list.

Underneath the announcements is a prolific attack-research operation, publishing CVE disclosures, exploit analysis, and threat guidance, that keeps Horizon3's name in front of practitioners between funding headlines.