(Picture courtesy VAST Knowledge)
It’s not day by day that you just see a brand new working system launched. However VAST Knowledge as we speak claims to have accomplished precisely that with the launch of what it’s calling a brand new working system for AI. The VAST AI Working System builds on the corporate’s present information and AI infrastructure developments and offers prospects the power to construct and run AI purposes utilizing their very own information.
VAST Knowledge says we want a brand new working system as a result of we’re on the “daybreak of a brand new computing paradigm.” Simply because the PC, cell, and cloud revolutions gave us new abstractions of complexity, the brand new agentic AI world calls for a brand new software program layer to democratize the usage of new {hardware}.
The period of deep studying has created the necessity for a brand new working system that may rework AI infrastructure into “easy to make use of and scalable machines,” VAST Co-founder Jeff Denworth says in a video launched as we speak.
“To us it’s develop into abundantly clear that the methods by which deep studying workloads are computing look completely nothing just like the enterprise methods of yesterday,” Denworth says. “We’re rapidly shifting to an age the place there will likely be tens of millions of GPUs and each conventional system will break underneath the load of this computing energy as a result of they had been by no means designed for this scale and this depth.”
(Picture courtesy VAST Knowledge)
The VAST crew has been constructing out its information platform since Denworth, Renen Hallak, and Shachar Fienblit, co-founded the corporate a couple of decade in the past. VAST took a clean-sheet method to constructing its platform round NVMe drives utilizing its Disaggregated and Shared Every thing, or DASE, idea.
The primary component of the brand new DASE method with VAST Knowledge Platform was the VAST DataStore, which gives massively scalable object and file storage for structured and unstructured information. Subsequent was the DataBase, which features as a desk retailer and gives information lakehouse performance just like Apache Iceberg. Then it launched the DataEngine, which gives the aptitude to execute features on information. Subsequent got here DataSpace, which gives a world namespace for storing, retrieving, and processing information from the cloud to the sting.
In October, the New York firm unveiled the InsightEngine, which makes use of Nvidia Inference Microservices (NIMs) from Nvidia and was the primary new utility designed to run atop the platform. In early 2025, the corporate introduced help for block storage and real-time occasion streaming by way of an Apache Kafka-compatible API. In April the corporate added a number of GenAI capabilities, together with help for vector search and retrieval; serverless triggers and features; and fine-grained entry management.
VAST Co-Founder Jeff Denworth
In the present day’s launch of the VAST Working System for AI hinges on the launch of another piece of the puzzle (what we surmise is a vital piece, since its inclusion permits VAST to have what it claims is a wholly new working system). That new part is named the VAST AgentEngine.
In response to VAST, the AgentEngine options an “AI agent device server” that permits brokers to invoke information, metadata, features, and execute Net searches, or name different brokers utilizing Mannequin Context Protocol (MCP), an rising commonplace for agentic AI created by Anthropic in late 2024. Brokers can assume completely different personas and sort out completely different duties utilizing completely different safety credentials, VAST says, whereas scheduling and tracing mechanisms present for environment friendly process completion and resiliency.
VAST says it would launch one open supply AI agent per thirty days to run on its AI OS. Among the examples of brokers embrace: a reasoning chatbot; an information engineering agent; a immediate engineer; an “agent agent” to assist work with and enhance different agent; a compliance agent; an editor agent; and a life sciences agent to assist with analysis.
(Picture courtesy VAST Knowledge)
Each technological revolution began with new {hardware} and finally led to new working methods, VAST CEO and Co-founder Hallak says. For example, the PC revolution gave us Home windows and Mac OS, the Web revolution gave us networking protocols and the Net browser, whereas cell computing gave us Android and iOS.
“It’s the working system, agnostic to the {hardware} beneath and purposes above, which defines these revolutions and create probably the most worth by democratizing new {hardware} and utility adoption,” Hallak says in a video posted as we speak. “AI will dwarf all the earlier computing brokers in each dimension and velocity.”
“Within the Nineteen Seventies, solely pc scientists may function computer systems. With Home windows everybody may,” he says. “In the present day we’re at that very same inflection level with respect to AI. Whoever builds the AI working system will democratize AI by making it easy, by making it safe, and by making it accessible to everybody.”
Associated Gadgets:
VAST Fleshes Out Knowledge Platform for Enterprise RAG Use Circumstances
VAST Knowledge Expands Platform With Block Storage And Actual-Time Occasion Streaming
VAST Appears to be like Inward, Outward for An AI Edge
