I have always been fascinated by what comes next.
Growing up, I watched technology reinvent itself again and again, from early gaming and the dot-com era to SaaS, cloud, and modern software development. I remember wondering when I would get the chance to help build what came next.
At the time, I was mostly watching it happen and wondering when I would get my own chance to build something that could be part of that future.
Eventually, I found that opportunity in cybersecurity. Over the years, I’ve had the chance to build products, research attacks, work on some of the hardest problems in cyber defense, and learn alongside incredibly talented engineers and researchers. Every generation of technology brought a new set of possibilities, but also a new set of security problems that had to be understood from the ground up.
Today, I believe we are entering the most important technology shift of the last decade. AI is changing how software is built, how infrastructure operates, and how production behaves.
For those of us building security products, that makes this an incredible moment.
And at Upwind, we intend to go big.
The future belongs to builders
Technology markets have always belonged to builders with strong foundations and the curiosity to keep learning as those foundations move underneath them.
AI is compressing development cycles, changing architectures, creating entirely new infrastructure layers, and giving software increasingly autonomous capabilities. Models are connecting to tools, identities, APIs, data and production systems. Agents are taking actions that required humans.
At Upwind, we have always believed that understanding risk starts with understanding what is actually happening at runtime. In the AI era, that matters more than ever.
AI does not live in one place, and neither can AI security. As the attack surface expands, so does the opportunity to rethink how we secure it.
That is why today I’m excited to announce two important steps forward: Upwind is acquiring Aegis, and we are launching the Upwind AI Lab.
Building teams is harder than building technology
One thing I have learned is that building a great cybersecurity team is one of the hardest problems in our industry. Technology can be bought, rewritten, or replaced. Finding people who share the same curiosity, intensity, and ambition is much harder.
When we met the Aegis team, the connection was immediate. We were asking the same questions, moving with the same urgency, and thinking about the future in the same way.
The culture matched. The ambition matched. The desire to build matched.
Basketball taught me that great teams are about more than individual talent. They are about people who compete, make each other better, and want to win together. Meeting Aegis felt like we were already playing on the same court.
Now, together with Upwind’s researchers, engineers, and product teams, they will help build the foundation of the Upwind AI Lab.
Introducing the Upwind AI Lab
The Upwind AI Lab is being built around a simple idea: give exceptional people the environment, infrastructure and freedom to work on security problems that are not only important today, but will define what the industry needs several years from now.
We want to combine the speed of a startup with Upwind’s engineering, production data, and security expertise. Some of this work will become product, some will become research, benchmarks, and open technical work, and some will simply help us learn faster.
Frontier research means working on problems before the answer is obvious, and occasionally before the market has even agreed on the question.
The Lab will initially focus on four major areas that together form an important part of Upwind’s broader AI strategy.
1. Security for AI
AI is becoming part of both the enterprise and the production space, spanning models, agents, GPUs, identities, APIs, data, SaaS platforms, and cloud infrastructure. We believe AI security cannot be treated as a separate layer because the real risk lives in how all of these components interact at runtime. The Lab will focus on securing the full AI stack with the same production-first approach Upwind brings to cloud security: understanding what is running, what it can access, how it behaves, and where meaningful risk actually exists.
2. Agentic Security
Security itself is becoming agentic. We have already started this journey with the Upwind Agentic Pack, and the next step is building agents that can investigate, correlate, reason, validate, and eventually take action with the right guardrails. The real challenge is not creating an agent that can talk about security, but giving it enough production context to understand what is happening across workloads, identities, networks, vulnerabilities, and data so it can make reliable security decisions.
3. Frontier Models, Research and Benchmarks
Cybersecurity contains some of the hardest reasoning problems in technology, from reconstructing attack paths to understanding exploitability across massive and constantly changing environments. We want to turn these challenges into serious research problems for frontier models by building new benchmarks, datasets, and evaluations based on real security scenarios. Our goal is to bring Upwind’s security knowledge into the broader research community and help push models from simply knowing security concepts to actually reasoning through complex security problems.
4. Advanced AI Systems
The Lab will also explore technologies that sit ahead of today’s product roadmap, including advanced retrieval architectures, security decision models, context and memory systems, GPU security, autonomous investigation, and new approaches to AI infrastructure security. Some of this work will become product, some will become research, and some will open entirely new directions, but the goal is the same: give strong builders the freedom and resources to work on difficult problems before they become obvious market categories.
This is our time to build
When I think back to the technology I grew up watching, what stays with me is how obvious each transformation looks in hindsight.
Of course the internet changed everything.
Of course software moved to the cloud.
Of course SaaS became the default.
But when you are standing at the beginning of a transition, nothing feels inevitable. There are competing ideas, unfinished technologies, and difficult problems everywhere.
That is where we are with AI today.
We don’t know exactly what production will look like five or ten years from now, but we know it will be more intelligent, more autonomous, and more connected than the systems we secure today. We also know that securing that world will require new architectures, new research, and a willingness to rethink assumptions that have defined cybersecurity for decades.
That uncertainty is not something I find intimidating, it is exactly what makes this moment exciting.
For Upwind, the acquisition of Aegis and the creation of the Upwind AI Lab represent another step in building for that future. We are bringing together exceptional people, and giving teams the space to attack problems whose solutions may define entirely new categories of security.
For me, it also feels personal. I grew up watching technology move forward and wondering what it would be like to help shape one of those defining moments.
Now we get to build through one.
Welcome to Upwind, Aegis.
Welcome to the Upwind AI Lab.
There is a lot to build.
Safe surfing.
