Horizon3.ai
Horizon3.ai's NodeZero attacks a company's own network the way an intruder would, chaining weak passwords, misconfigurations and unpatched systems into proof of what an attacker could reach, and can rerun the test as often as the team wants.
CyberReviewed Editorial Team · Updated · horizon3.ai
Company details
- Score
- 7.8/10Ranked 4 of 12
- Category
- Penetration testing
- Founded
- 2019
- Headquarters
- San Francisco, California
- Leadership
- Snehal Antani, co-founder and CEO
- Funding disclosed
- $178.5 million, including a $100 million Series D in 2025
- Pricing
- From $25,000 per year for 500 assets (AWS Marketplace)
- Best fit
- IT and security teams that want frequent internal network pentests
Our take
Horizon3.ai is worth buying for organizations that want to test their internal network as an attacker would every month or every week instead of once a year. Teams whose risk sits mostly in custom web applications will also need human application testing.
Scores by criterion
| Criterion | Score |
|---|---|
| Pricing transparency | 8 |
| Product scope and fit | 8 |
| Company durability | 8 |
| Evidence of adoption | 8 |
| Ease of adoption | 7 |
| Overall score (average of the five) | 7.8 |
Each criterion is defined on our methodology page.
Autonomous attacks on production
NodeZero, Horizon3.ai's platform, runs autonomous penetration tests against production environments. It chains exploitable vulnerabilities, misconfigurations, default settings, weak controls and exposed credentials into attack paths and provides proof of exploitation so teams can see how an attacker would move.1 Add-on modules include deception tripwires, rapid alerts on newly disclosed threats and trend reporting, and a separate web application testing option is sold alongside.2
Package prices on AWS Marketplace
Horizon3.ai's AWS Marketplace listing gives a price for each package. Twelve-month packages for 500 assets cost $25,000 for Core, $32,500 for Pro and $42,500 for Elite; a one-time Flex test across 1,000 assets costs $15,000; and the web application product is listed at $10,000.2 For a mid-size company that would otherwise pay for one or two consultant-led internal tests a year, the Core package allows unlimited recurring tests within its asset limit.
Series D and federal plans
Horizon3.ai was founded in 2019 and is based in San Francisco, California; Snehal Antani is co-founder and CEO.1 In June 2025 it raised a $100 million Series D led by NEA, with SignalFire, Craft Ventures and 9Yards Capital, bringing total funding to $178.5 million, and said more than 3,000 organizations used its platform.13 Lila Tretikov, a partner at NEA and former deputy CTO of Microsoft, joined its board.3 The company said it would use the money to grow its partner ecosystem, add web application pentesting, vulnerability management and precision defense capabilities, and expand adoption within federal agencies.1
Human pentests and scanners
Horizon3.ai competes with other automated attack and breach simulation tools and with traditional internal pentests. Cobalt's human testers are stronger on application logic; NodeZero can test infrastructure far more often and at larger scale. Intruder scans for known issues, while NodeZero tries to exploit them and shows the resulting path.
Production limits and compliance reports
Organizations that cannot accept automated exploitation in production, even when controlled, can run NodeZero against a staging network first. Some compliance regimes expect a signed report from an individual human tester, which an autonomous test does not produce.
What customers say
Quoted word for word from public reviews and posts on PeerSpot and Reddit, retrieved October 21, 2026. Each quote links to the review or post it comes from.
Where it helps
The NodeZero Platform by Horizon3.ai has positively impacted my organization by giving us a better continuous picture of our security posture, what's exploitable, and what can be used against the organization.
The automated updating of scans has been a huge value-add for us; we're a small cybersecurity team, so speed to value is the biggest thing to multiply the force of our small team, saving a lot of man-hours.
horizon3.ai is fairly good though, it will actually give you proof that it was able to exploit a system and how.
Where it falls short
We did hundreds of tests, so that is why we did not continue, as it was very expensive for a very low yield.
