Microsoft Discovery is an enterprise-grade research and development (R&D) platform designed to accelerate scientific and engineering innovation by combining artificial intelligence (AI), high-performance computing (HPC), and advanced knowledge management. It enables organizations to automate complex research workflows, integrate proprietary and public data, and orchestrate multi-agent systems ...
An overview of Microsoft Discovery's Service Architecture A workspace is the top-level ARM resource in Microsoft Discovery. It acts as the shared infrastructure boundary for all projects within it, providing: The networking configuration (virtual network and subnets) used by all child resources The supercomputer and chat model deployments shared across projects The access control layer for ...
Set up Microsoft Discovery infrastructure by creating a supercomputer, workspace, project, agents, investigations, and then run your first AI-powered scientific investigation.
Step-by-step guide to setting up investigations with the Discovery Engine, from creating your first task to monitoring an autonomous research investigation.
A Supercomputer is a managed compute cluster in Microsoft Discovery. NodePools provide the underlying virtual machines (VMs) that run workloads on the Supercomputer. You can attach multiple Node Pools, each with different VM types and scaling limits, to a single Supercomputer. This article is scoped only to Supercomputer and NodePool creation. Prerequisites such as networking, identities, and ...
This quickstart shows you how to deploy the prerequisite infrastructure for Microsoft Discovery using Bicep.
Learn how tasks work in Microsoft Discovery. Understand task structure, the status lifecycle, dependencies, and how cognition uses tasks to organize and execute research.
Learn how to write agent instructions and user prompts for Microsoft Discovery agents to get accurate, well-structured responses for scientific research tasks.