Last fall, the investment community received an unusual and high-priority message from Roy Reznik, one of the co-founders of cybersecurity unicorn Wiz. Reznik offered to introduce a founding team that had just walked out the doors of Microsoft. Because it was the first and only time Reznik had ever reached out to recommend a founding team, investors immediately paid close attention, recognizing that someone of his background and technical acumen would not lightly endorse a group of departing engineers.
The team Reznik wanted to put in front of investors had spent more than a decade quietly building Kusto, a powerful data engine that originated as a small incubation project deep inside Microsoft. Named after the legendary oceanographer Jacques Cousteau because it was specifically engineered for diving into deep oceans of data, Kusto was designed from the ground up to retrieve answers from enormous volumes of fast-arriving information—such as system logs, telemetry feeds, and security events—in a matter of seconds rather than hours. Beyond its raw speed, Kusto was built on a query language that was simple enough for everyday users rather than just specialized big-data engineers.
The technology worked so exceptionally well that it ultimately became the foundational data engine running underneath several of Microsoft’s flagship products, including Microsoft Sentinel, Microsoft Defender, and Azure Monitor, before eventually being packaged and offered to enterprise customers as Azure Data Explorer. Security operations teams, in particular, fell in love with its capabilities, which was precisely how Roy Reznik became one of its most vocal and passionate champions. When investors subsequently spoke with enterprise customers ranging from massive global corporations to fast-growing startups, they heard the exact same two themes echoed consistently: first, that Kusto was an absolute game-changer in data technology, and second, that the engineering team behind it was among the most talented and collaborative they had ever worked with in their professional lives.
That elite team consisted of Avner Aharoni, Evgeney Ryzhyk, Alexander Sloutsky, and Yoni Leibowitz. Each member brought a distinct, highly specialized set of skills to the table. Evgeney actually invented the original Kusto engine and its accompanying query language, and then impressively rebuilt the architecture three distinct times from the ground up, making it progressively faster and more cost-efficient with every single iteration. Alexander possessed a rare talent for transforming half-formed, abstract product concepts into fully polished, finished features seemingly overnight. Yoni took charge of the architectural rigor, ensuring that every complex design was thoroughly thought through and vetted before a single line of code was ever written. Avner, who accumulated a quarter-century of experience at Microsoft, served as the driving force who hustled Kusto from an internal incubation project into a globally beloved commercial product. Their new startup, Edelbyte, builds directly upon this rich history of technical success, and investors have proudly led their seed round to establish a formal partnership.
In the world of B2B technology, this trajectory represents a proven and highly successful playbook: a brilliant team develops an innovative solution inside the controlled confines of a massive technology corporation, validates its market utility, and eventually spins out to build an independent company around the core technology. This is the exact origin story behind prominent venture-backed portfolio companies like Confluent, whose founders originally created the Apache Kafka data streaming platform while working at LinkedIn, and Temporal, whose founders built the Cadence workflow engine while at Uber. Edelbyte is now stepping out onto that very same path, looking to repeat the formula on a new technological frontier.
However, Edelbyte is not simply a fourth iteration or a direct rebuild of Kusto for its own sake. Instead, it represents a fundamental rethinking from first principles of what a modern data platform actually needs to look like in the age of artificial intelligence. With every major paradigm shift in computing history, the fundamental way software is built changes, and the underlying data infrastructure supporting it must evolve in tandem. Traditional data platforms were architecturally designed around human users, operating under the assumption that a person would type queries manually. Because of this human-centric design, there has always been a natural, physical limit to the number of questions a single person can formulate and type into a system in a single day.
That inherent limitation made various historical trade-offs much easier for developers and enterprise buyers to tolerate. These compromises included retaining less historical data so that queries would stay fast, limiting the number of analytical questions asked so that the monthly cloud bill remained reasonable, and building the product tightly coupled to whichever specific cloud environment the database happened to run on.
Today, however, the landscape has fundamentally shifted with the rise of autonomous AI agents. An AI agent investigating a technical problem or security incident behaves less like a single human user and more like an entire team of deeply curious analysts who never get tired, testing dozens of multiple hypotheses concurrently. While a human analyst might run a single query, an autonomous agent might fire off fifty related queries in rapid succession to trace an anomaly. At that massive scale of automated inquiry, traditional rationing mechanisms and legacy trade-offs quickly constrain the true, transformative promise of artificial intelligence. An AI agent is ultimately only as useful as the questions it can actually afford to ask, and what it can afford is dictated entirely by the economics and performance of the underlying data layer.
To solve this mismatch, Edelbyte keeps detailed historical data fully queryable at interactive speeds, scales seamlessly as both data volumes and automated query demands grow exponentially, and provides people, traditional applications, and AI agents with unified, governed access to the exact same dataset. Furthermore, the platform brings search and analytics capabilities—which historically have had to be stitched together from separate, disparate systems at significant financial cost and operational complexity—into a single, cohesive platform. Crucially, it is designed to run wherever an organization needs it to operate: whether deployed in the public cloud, on-premises infrastructure, or managed and self-hosted environments.
The resulting architecture is precisely what heavy AI workloads require: an environment where organizations no longer need to ration curiosity, regardless of whether that curiosity originates from a human engineer or an autonomous AI agent. Early enterprise customers are already reporting dramatic increases in operational efficiency, including instances where query volumes have increased tenfold without incurring any corresponding increase in their overall technology budget.
Within the technology sector, data platforms have consistently proven to be among the most durable and lucrative business models because data possesses a unique gravitational pull; once vast amounts of information accumulate in a single repository, it rarely moves elsewhere. Every major technological era produces its own defining winners, and the prevailing belief is that the artificial intelligence era will be won by the platform that empowers builders, along with the autonomous agents working on their behalf, to ask as many questions as they desire without ever having to worry about query latency or the final monthly bill. Edelbyte is currently constructing precisely that kind of platform, and its backers are eager to support the journey ahead.

Comments (0)