Seven months ago, we introduced Fabrix with a clear belief: Identity security needs to be rebuilt for the AI era. We started with the foundation: making identity security AI-native.
For decades, identity and access management has relied on a simple approach: define access in advance, enforce it consistently, and review what went wrong after the fact. That model assumed relatively predictable behavior, static permissions and stable roles, and human-driven access patterns.
Those assumptions no longer hold.
So we set out to rethink Identity Security from the ground up. We built an AI-native system designed to understand identities, permissions, and access patterns in full context. One that could continuously analyze who has access to what, how it is used, and where risk actually exists. Not through manual processes, but through AI.
This allowed us to solve some of the hardest problems in identity. Making sense of identity sprawl. Automating access reviews. Moving toward least privilege in environments where both human and non-human identities are growing exponentially.
But as we went deeper, a more fundamental issue became clear: even the most intelligent system, operating at admin time, is not enough.
We are now operating in environments where non-human identities outnumber humans, AI agents act autonomously, and access decisions need to happen in milliseconds. Policies defined ahead of time cannot handle this speed and complexity. Periodic reviews, even if automated, are irrelevant when risk materializes in seconds.
This is not a scaling problem. It is a paradigm problem.
From predefined access to real-time decisions
This led us to a simple but fundamental shift that shaped our vision. If I had to describe it in one sentence, it would be this: build an AI-native system that can make access decisions and enforce them in real time, based on full context, at the exact moment they are needed.
Because the real question is no longer: “What access should this identity have?”
It is: “What is this identity trying to do right now, and should it be allowed?”
Answering that requires more than identity and roles. It requires understanding behavior, intent, timing, and business context. Continuously.
As Andrej Karpathy has pointed out, AI systems are only as good as the context they operate on. In identity security, that context is everything. Who the identity belongs to, what it’s trying to do, what it has done before, what it should be allowed to do, and under what conditions. Without that, even the most advanced AI will make the wrong decision.
This is why we built an identity-centric AI decisioning engine, powered by a living identity knowledge graph that brings together identity data, permissions, activity, policy, assets, and organizational context to evaluate access in real time.
In practice, this means moving away from static authorization toward continuous, real-time decisioning. Supporting Just-In-Time access by default. Adapting to changing behavior as it happens, not after the fact.
But as we built, we understood a simple truth.
Decisioning without runtime enforcement is not enough
Making the right decision is critical. But on its own, it doesn’t stop threats.
If a decision cannot be enforced at the exact point where access happens, in real time, it becomes advisory. And with the growing speed of AI agents and modern attacks, that’s just too late.
This is where many security solutions fall short. They analyze. They recommend. They alert. But they are not embedded in the execution path.
To actually secure identity and access, decisioning and enforcement must operate as one, at the right time: runtime.
This is why the synergy with Silverfort was clear and immediate.
Why Silverfort
Silverfort has built what we believe is the other missing layer in identity security: runtime enforcement embedded directly in the authentication and access flow. Their Runtime Access Protection (RAP) technology enables inline enforcement without changing the infrastructure or rewriting applications. A unique architecture, built on native integrations and deep ecosystem partnerships, creates multiple enforcement points across human users, non-human identities, and AI agents – both on-prem and in the cloud.
This is not just an architectural advantage. It’s what makes real, effective identity security possible.
By joining Silverfort, we can combine our AI-driven decisioning engine with a proven enforcement layer that already operates at scale in more than 1,000 companies, including some of the largest enterprises in the world.
What we’re building together
Together, we are building a new model for identity security in the AI era.
An autonomous identity security platform that continuously evaluates and enforces access decisions at runtime. One that can operate at the speed, scale, and unpredictability of modern environments, including humans, machines and AI agents.
At its core, it answers a simple but critical question: what is this identity trying to access, and should it be allowed in this specific instance? And it answers it instantly, using AI and real-time context, at a speed and scale humans can no longer handle.
We’ve always believed that what we’re building could become an AI “brain” for identity. A system present in every access decision, across every environment, for every type of identity. With Silverfort’s ability to enforce decisions inline everywhere, that vision becomes real.
Why now
AI models, and with them AI-driven threats, are evolving rapidly. This creates a narrow window. Identity security is about to change fundamentally, and it will happen soon, measured in quarters, not years. We believe that the identity security models that organizations relied on for the past years and even decades will soon become unable to keep up with this new speed, scale, and autonomy of AI-driven environments. The gap is already forming, and identity security must evolve urgently. There is no time to waste.
This decision is about speed and velocity. In the AI era, that matters more than ever before. Not just as a technical requirement, but also as a strategic one. The ability to bring innovation to customers at scale, and to learn from real-world environments is what defines whether a technology can become foundational. Joining Silverfort allows us to do exactly that. For us, this decision is about time-to-market and impact. To bring what we built to some of the largest organizations in the world, where identity complexity is highest, and the need is immediate. And to do it now.
That is a unique opportunity, and one we are very excited to take.
Looking ahead
We are thrilled to continue building and innovating as part of Silverfort, and to help define what identity security will look like in the AI era.
I want to thank the people who made this journey possible.
To our investors, thank you for your belief and support from day one.
To our customers, thank you for trusting us early and helping shape the product through real use and real feedback.
Most importantly, to the Fabrix team. This is yours. What we built in such a short time reflects your talent, ownership, professionalism, and ambition. I’m proud of what we’ve done together and excited for what comes next.
And to the Silverfort team, thank you for the trust and for sharing the same vision. It’s rare to find this level of alignment on both the problem and how to solve it.
We’re just getting started. Let’s build the future of autonomous Identity Security together!
