Reflectiz Unveils AI‑Driven Pentesting Platform Claiming Tenfold Coverage Boost

Boston‑based cybersecurity firm Reflectiz announced the launch of its new Agentic Pentesting service on September 8, 2026, positioning the offering as a dramatic upgrade to traditional web vulnerability testing.

The service deploys a dedicated team of artificial‑intelligence agents that automatically discover, probe, and confirm weaknesses across a website. By drawing on the existing context of the target site—such as known page structures, scripts, and prior findings—the agents can filter out false positives and focus on genuine risks, a process the company says trims the time needed for remediation.

Reflectiz describes the approach as delivering up to ten times the coverage of conventional penetration tests. The claim rests on the agents’ ability to execute parallel attack vectors and maintain persistent awareness of a site’s evolving attack surface, something that manual testing teams often struggle to replicate within limited engagement windows.

The launch is part of Reflectiz’s broader continuous web exposure management strategy, which seeks to shift vulnerability assessment from periodic, point‑in‑time checks to an ongoing, automated surveillance model. In theory, the shift could help organizations stay ahead of attackers who constantly probe public‑facing applications for new flaws.

Industry observers noted the announcement after GBHackers first reported the development, and CyberNewswire carried the story in its specialized feed. Analysts highlighted the growing interest in AI‑assisted security tools, while also cautioning that automated testing must still be complemented by human expertise to interpret findings and prioritize fixes.

Reflectiz did not disclose pricing or rollout timelines, but indicated that early adopters will receive integration support to embed the agents into existing security operations. The company’s next steps appear to involve expanding the agent library to cover emerging web technologies and refining the noise‑reduction algorithms based on real‑world feedback.