Partnering with Edelbyte: Data Infrastructure That Keeps Up With AI's Curiosity

Last fall we got a message from Roy Reznik, one of the co-founders of Wiz, offering to introduce us to a team that had just left Microsoft. This was the only introduction to a founding team we’d ever received from Roy, and we paid attention.

The people Roy wanted us to meet had built Kusto, a data engine that started more than a decade ago as a small incubation project inside Microsoft. Named after the oceanographer Jacques Cousteau because it was built for diving into oceans of data, Kusto was designed to get answers from enormous volumes of fast-arriving data, such as logs, telemetry and security events, in seconds rather than hours. It ran on a query language simple enough that anyone, not just big-data engineers, could use it. And it worked so well that it became the data engine underneath Microsoft Sentinel, Microsoft Defender and Azure Monitor, and was eventually offered to customers as Azure Data Explorer. Security teams in particular loved it, which is how Roy became one of its biggest champions. When we spoke with customers, large enterprises and fast-growing startups alike told us two things: it was game-changing data technology, and the team behind it was among the best they’d ever worked with.

Avner Aharoni, Evgeney Ryzhyk, Alexander Sloutsky and Yoni Leibowitz were that team. Evgeney invented the Kusto engine and its query language, then rebuilt it three times, making it faster and cheaper with each iteration. Alexander turned half-formed ideas into finished features seemingly overnight, and Yoni made sure every complex design was fully thought through, starting before a line of code was ever written. Avner, who spent a quarter century at Microsoft, hustled Kusto from project to a globally beloved product. Their new company, Edelbyte, builds on this success. We are proud to have led their seed round and to be their partners.

In B2B technology, this is a proven playbook: a team develops a solution inside a large company and then splits off to build it independently. It’s the origin story of Sequoia portfolio companies such as Confluent, whose founders created Kafka at LinkedIn, and Temporal, whose founders built Cadence at Uber. Edelbyte now follows the same path.

But Edelbyte is not simply Kusto rebuilt for a fourth time; it is a rethinking, from first principles, of what a data platform should look like in the age of AI. With every major shift in computing, the way software is built changes, and the data infrastructure underneath it has to change too. Data platforms were designed around human users, and there is a natural limit on the number of questions one person can type in a day. That made the trade-offs, such as keeping less history so queries stay fast, asking fewer questions so the bill stays reasonable, and building the product inside whichever cloud the database happens to run in, easier to tolerate.

But an AI agent investigating a problem is like a whole team of curious analysts who never get tired, testing multiple hypotheses concurrently. One agent might fire off fifty queries where a person would run one. And at that scale, rationing and trade-offs quickly constrain the true promise of AI. An agent is only as useful as the questions it can afford to ask, and what it can afford is decided in the data layer.

Edelbyte keeps detailed history queryable at interactive speed, scales as data volumes and query demand grow, and gives people, applications and AI agents governed access to the same data. It brings search and analytics, which have historically been stitched together from separate systems at significant cost and complexity, together in a single platform. And it runs wherever you need it: in the cloud or on premises, managed or self-hosted.

The result is exactly what an AI-heavy workload needs: no more rationing curiosity, whether it comes from a person or an agent. Already, customers are seeing dramatic increases in efficiency, including tenfold increases in queries with no increase in budget.

Data platforms are among the most durable and lucrative businesses in technology, because data has gravity; once it accumulates in one place it rarely moves. Each era produces its own winners, and we believe the AI era will be won by the platform that lets builders, and the agents working for them, ask as many questions as they want without worrying about the speed or the bill. Edelbyte is building just such a platform, and we are thrilled to be part of the journey.