RemoteThreat — Building the OS of Offensive Cyber

Today, RemoteThreat is emerging from stealth, and we’re thrilled to lead its $7M pre-seed alongside DataTribe.

RemoteThreat is building an Offensive Cyber Operations (O/C/O) platform that unifies advanced tooling, stealth-focused tradecraft, command-and-control, and governed AI. The long-term ambition is to become the OS for offensive cyber, giving red teams and government mission teams greater speed and control as operations become increasingly agent-driven.

Why Now — The Dawn of Cyber-Superintelligence

AI cyber capabilities are improving at an alarming pace. In just two years, frontier models have gone from solving a small fraction of professional cybersecurity challenges on benchmarks like Cybench to completing the vast majority. They are finding vulnerabilities, developing exploits, and executing increasingly long portions of an attack chain autonomously, while faster inference, smaller models, and better multi-agent coordination make it possible to deploy that capability at greater scale.

As those capabilities compound, cyber-superintelligence becomes real. Humans increasingly define the objective while networks of agents map systems, discover vulnerabilities, establish access, and adapt at a speed and scale no human team can match.

For U.S. government agencies and enterprises, the implications are serious. Offensive AI lowers the cost of sophisticated cyber operations while allowing adversaries to run more of them simultaneously. Defending against that world will require trusted teams to field comparable capabilities on both offense and defense without sacrificing the control and judgment that real missions demand. As more cyber capability moves into the model, the systems around the model become more important, not less.

The Missing Context Layer

AI can increasingly execute an attack, but conducting a real operation requires a different kind of judgment. In the Hugging Face intrusion, agents completed an end-to-end attack chain while generating roughly 17,600 attacker actions. An experienced operator would take a far narrower path because every unnecessary probe, failed attempt, or redundant action increases the probability of detection.

Experienced hackers know when to act, what conditions need to be true first, how long to wait, which paths are becoming too risky, and when to change course. Some of that tradecraft can be codified into TTPs and mission state, while some must be embedded directly into the systems guiding agent behavior. Better models can execute more techniques, but they do not automatically inherit the judgment required to use them well.

RemoteThreat is building the operating environment around those models. It is starting with semi-autonomous operations, using the O/C/O platform to keep expert operators in control while agents take on more of the work. That control layer is what allows autonomy to expand as the system proves itself in real operations.

We believe this is the last mile of the cyber frontier, where capable hacking models become systems that can operate with the judgment, stealth, and control of expert operators.

Built by Hackers

Building that layer requires people who understand offensive operations from the inside. Chris Thompson (“CT”) and Shawn Jones have spent their careers breaking systems and building the infrastructure behind complex cyber operations. They have assembled a team of hackers, researchers, and engineers from places including IBM X-Force, Mandiant, SpecterOps, Dreadnode, Bugcrowd, Microsoft, defense contractors, and government organizations.

There is a countercultural streak to the company that feels native to the best traditions of hacking. The team is skeptical of received wisdom, deeply technical, and obsessed with whether something works. CT and Shawn have extended that ethos through Offensive AI Con, which brings together researchers and practitioners exploring autonomous security testing, multi-agent systems, and offensive AI. Vijay Bolina, Dave Aitel, Joshua Saxe, Chompie, and others around the company reinforce a community that is unusually close to the technical frontier.

We invested in the team because we believe offensive cyber is entering a new era. Models are becoming capable hackers, but the systems required to turn that capability into trusted, controlled operations are still being built. RemoteThreat is building the control layer that lets increasingly capable agents operate with expert tradecraft.

We’re honored to partner with CT, Shawn, and the rest of the team as they build the operating system for the next era of offensive cyber. The team is hiring and working with a select group of red teams and government mission partners.