Metalware, Inc.

We are excited to partner with the Air Force Research Laboratory and AFRL/RI Information Institute to advance the frontier of embedded systems cybersecurity and binary analysis. Metalware, Inc. is working with Zion Leonahenahe Basque, incoming Assistant Professor at the University of Georgia, to research Adaptive LLM-Driven Decompilation for Robust Binary Analysis. Zion was previously co-captain of Shellphish, one of the world's oldest ethical hacking groups, and led their autonomous patching group to win $3M at Defense Advanced Research Projects Agency (DARPA) AI Cyber Challenge (AIxCC). He recently completed his PhD under Ruoyu "Fish" Wang and Yan Shoshitaishvili at Arizona State University, and is also a core developer of the widely-used angr decompiler. We look forward to continuing to bring technology to the Department of War in support of the warfighter.

Everyone is talking about AI. Very few understand where it actually matters. Here’s one place it does: Cybersecurity. The firmware running inside your vehicle. Your drone. Your industrial controller. The embedded code sitting at the exact boundary between software and the physical world, and the vulnerabilities living inside it that manual testing will never catch fast enough. AI doesn’t just belong in this conversation. It may be the only way we win it. Ryan Chow, Co-Founder & CEO, and Zion Leonahenahe Basque, Research Fellow, both of Metalware, Inc., take the CPS3 stage June 16 at 1:00 PM for a session that cuts past every AI headline you’ve seen this year: Towards Automating Firmware Testing — and How AI Can Help The teams that figure out where AI actually performs in cyber-physical defense won’t be talking about it on a panel two years from now. They’ll already have won. CPS3. June 16–17 Oakland University In the Global Epicenter of Mobility. This is where AI gets operational.

Metalware, Inc. has achieved Awardable status in the DoD Platform One (P1) Solutions Marketplace for Cybersecurity and DevSecOps. This means government teams can rapidly acquire and deploy Metalware, bringing autonomous firmware testing directly into operational environments to secure uncrewed autonomous systems, weapons platforms, tactical communications, ISR/EW systems, OT / critical infrastructure, and other mission-critical cyber-physical assets. Your contracting team can request the P1SM assessment materials, use the marketplace as the competed basis, work with us on scope and pricing, and award via the appropriate CSO/BAA/OT/FAR instrument without running a new solicitation. Government users can access the solution at the link in the comments 👇

Embedded systems cybersecurity is fundamentally different from securing desktop software, webapps, and other general-purpose software. And despite the hype, today's LLMs don't solve every problem on their own. At Metalware, Inc., we pair deep expertise in embedded development, vulnerability research, and reverse engineering with the latest in AI to push what's possible. In our latest blog post, we share how we automate peripheral map discovery on monolithic firmware and compare how LLMs stack up. 👇 Link to blog post in comments

The big theme at RSAC Conference this year is agentic AI, on both sides of the fight. Researchers showed attack chains that are nearly fully automated, from recon to exploitation. Meanwhile, defenders are building autonomous agents to keep pace. As a RSAC Innovation Sandbox Top 10 Finalist in 2025, we've been thinking about this in the context of firmware analysis at Metalware, Inc. So we wrote up how we're using agents at Metalware to automate the hardest parts of firmware fuzzing: bootstrapping emulators, iterating on device configurations, and classifying crashes. All without source code or hardware. If you're shipping embedded systems, this is the kind of testing that should be running in your CI/CD pipeline every night. DM me if you'd like to meet for coffee at #RSAC2026 this week. Read our full blog post in the comments below 👇