Assail Launches from Stealth with Ares, Autonomous AI Agents for Continuous Penetration Testing Across APIs, Mobile, and Web Infrastructure

Company raises $250K pre-seed led by Squared Circle Ventures, secures $100K in AWS credits, joins NVIDIA Inception, and drives 264 early-access registrations following ISC2 Security Congress keynote debut

LAS VEGAS, Jan. 13, 2026 /PRNewswire/ -- Assail, Inc., a cybersecurity company building autonomous AI agents for API-first offensive security testing, today announced it is launching from stealth with Ares, an agentic offensive AI platform for continuous, machine-speed penetration testing across APIs, with coverage extending to web and mobile applications. Ares is designed to compress validated vulnerability discovery and remediation windows from weeks to minutes, helping organizations reduce risk before attackers exploit exposed APIs and modern application infrastructure.

Assail also announced it has closed a $250,000 pre-seed round led by Squared Circle Ventures, and secured $100,000 in AWS cloud credits to support the platform's compute-intensive development and deployment. In addition, as of January 6, 2026, Assail has been accepted into the NVIDIA Inception program, expanding Assail's access to NVIDIA's startup network and increasing visibility with the venture and investor ecosystem surrounding AI infrastructure.

"APIs are the invisible plumbing of everything — every bank, every hospital, every power grid," said Alissa Knight, CEO and Chief AI Officer of Assail. "I've spent my career proving how fragile that plumbing really is. Now we've built something that can find fractures faster than any human red team — including me."

"APIs are the invisible plumbing of our digital economy — and increasingly so in the AI era," said Karl Mattson, Managing Partner at Squared Circle Ventures. "Alissa Knight possesses the rare offensive tradecraft required to harden the world's critical API stack against the unknown."

Ares was created by Knight, a globally recognized hacker and two-time cybersecurity founder with successful exits. Her tools are featured in The Mob Museum's global cybercrime exhibit, and her published research includes offensive work across APIs and modern connected systems, including her book published in 2020 by Wiley & Sons, Hacking Connected Cars: Tactics, Techniques, and Procedures.

Traditional application security tools rely heavily on known signatures and predictable detection logic. Ares operates differently. Powered by a proprietary 14-billion parameter model fine-tuned on offensive tactics, techniques, and procedures, Ares can deploy up to 100 coordinated AI agents per target, communicating, sharing context, and executing multi-step attack chains like a full human red team, but at machine speed.

"What sets Ares apart is its operational intelligence," said Eduardo Ortiz Ramirez, CEO and Co-Founder of WATR Inc. "The autonomous agents discovered vulnerabilities in our mobile apps that would have taken our team weeks to identify manually. The mobile app and API exploitation capabilities are exceptional, and the closed-loop learning means the system gets smarter with each engagement — delivering compounding value on our security investment. This is applied AI solving real security challenges, not theoretical capabilities."

Ares also generates its own synthetic training data, effectively wargaming against itself to discover vulnerabilities that do not yet appear in conventional attack libraries.

"Most security models are trained on yesterday's attack techniques — and those labels start decaying the moment attackers adapt," Knight said. "The problem isn't that defenders lack data; it's that static, human-curated tactics quickly become a record of what we already understand. Ares is trained differently: it co-evolves. We pit an adversary simulator against a breacher in a continuous loop, generating new challenges at the edge of what the system can currently solve, and validating outcomes in sandboxed environments. That's how you move from pattern matching to discovery."

Alongside today's launch, Assail has open-sourced the Ares Docker Agent, now available on GitHub and Docker Hub, enabling enterprises to deploy Ares behind the firewall to hunt shadow APIs and internal applications invisible to external scans.

"Learning about Ares and its offensive agentic AI capabilities motivated me to become an early adopter so my team can start validating it in production environments," said Jeffery Buelt, VP of Security at Branch. "With Ares, we expect to increase the frequency of penetration testing cycles, expand coverage to more applications, and verify remediation faster. It's a meaningful advantage — stronger results, higher velocity, and lower risk."

"The biggest risk in modern enterprises isn't what's already cataloged in a signature library — it's what's hiding in the seams: undocumented APIs, internal services, and business logic that was never threat-modeled," Knight said. "Open-sourcing the Ares Docker Agent is how we bring this capability behind the firewall, where most of the real attack surface lives. The goal is simple: give teams a continuous, evidence-backed way to test the paths attackers actually take, not the ones a dataset happens to contain."

Knight unveiled Ares as a keynote speaker at ISC2 Security Congress 2025, and within days, more than 264 companies registered for early access. Assail reports pilots are currently underway with Fortune 100 manufacturers, U.S. state and federal agencies, the U.S. Department of War, financial services, and leading children's and research hospitals.

"In a market where many 'AI security' offerings feel like variations on the same theme, Ares stood out with a genuinely different approach," said Alex Roman, Manager of Information Security, Rady Children's Health. "What impressed me most was the clarity of its offense-led strategy—an agentic system purpose-built for authorized mobile, web, and API security testing—aimed at delivering high-signal, actionable findings that help teams harden real-world attack paths, not just AI for AI's sake."

"We're looking forward to working with Assail in 2026 and exploring how Ares can shrink exposure windows and deliver continuous, live-fire security testing across our web applications, mobile environments, and API attack surface," said David Wimer, Manager of the Vulnerability Management team at Caterpillar Inc.

Ares will be generally available on February 1, 2026, with expanded enterprise deployment options and additional agent capabilities.

About Assail, Inc.
Assail builds autonomous AI agents for API-first offensive security testing. Headquartered in Las Vegas, the company's flagship platform, Ares, delivers continuous, machine-speed penetration testing across APIs, with coverage extending to web applications and mobile environments. Ares combines agentic reasoning with safe-by-design "live-fire" simulation, progressive autonomy, and full audit trails to help organizations identify exploitable risk faster and shrink exposure windows from weeks to minutes.

Resources:
Ares Docker Agent (GitHub): github.com/assailai/ares-agent
Ares Docker Agent (Docker Hub): hub.docker.com/r/assailai/ares-agent

SOURCE Assail, Inc.