Company: Pentrify.cloud
Founder: Viktor Bulanek
Website: https://www.penetrify.cloud
About Pentrify.cloud
Penetrify started in October 2025 with a simple observation. Most companies ship new code every week, but test its security once a year, if at all. A manual penetration test costs thousands and takes weeks to schedule, so small and mid sized teams either skip it or live with eleven months of untested changes. We wanted to make real offensive testing something a team can run on every release, at a price a startup can say yes to.
Penetrify uses autonomous AI agents that test web applications and APIs the way a real attacker would, chaining weaknesses together to prove impact. They can sign in with test accounts, including accounts protected by two factor authentication, to reach the parts of an app only logged in users see. Penetrify also analyses Android and iOS apps and checks email security, from SPF, DKIM and DMARC to real email flows. Each confirmed finding comes with a proof of concept and clear guidance on the fix. Before any test starts, the customer has to prove they own the target, so the agents only touch systems they are authorised to test. A first test starts at 29 USD.
The challenges were real. The first was cost. Early full scans burned through model usage fast, which made them impossible to sell at small team prices. We rebuilt how the agents work, with hard budget limits per scan, caps on how many helper agents can spawn, and lighter models for simpler tasks. The second was trust in a clean report. When an AI model declined part of a task, that part of the test could quietly stop while the report still said nothing was found. A silent gap that looks like a pass is the most dangerous result a security tool can give. We now recover from those refusals and flag anything that did not finish. The third was agents settling for the first issue they found. We added checks that require evidence before an agent may call the job done, which raised our solve rate on public vulnerability benchmarks substantially.
On impact, we chose transparency over marketing claims. Our engine solved all 104 challenges of the public XBOW validation benchmark, and we published every log so anyone can check how each result was reached. In an industry where AI security claims are hard to verify, we think open evidence should be the standard. Our founder has also shared practical lessons on running autonomous agents safely in outlets including CSO Online, Computerworld and The Register.
Today Penetrify serves paying customers, offers a free security grader that gives any website owner a first look at their exposure, and is piloting a white label program so managed service providers and security consultancies can offer continuous testing under their own brand. Our goal is to move the industry from an annual pentest to testing that keeps pace with how fast software actually changes.
Want to find out what TechRound can do for your business?
