Microsoft HQ in Israel

Microsoft puts Israeli R&D at the center of its next generation of AI tools

Microsoft unveiled Microsoft IQ and Project Perception at Ignite 2026, two technologies developed with major contributions from its Israeli R&D center. The announcements reflect a broader push to turn AI agents into autonomous participants in the workplace, with access to corporate knowledge and the ability to carry out tasks.

Developments from Microsoft’s R&D center in Israel took center stage in several of the company’s major announcements at its Ignite 2026 conference, which is taking place in San Francisco. Among the most notable were Microsoft IQ, a technology that connects AI agents to the organizational knowledge they need to perform tasks, and Project Perception, a cybersecurity system in which teams of AI agents simulate attacks, detect threats and address vulnerabilities. Both were developed by engineers at Microsoft’s Israeli R&D center in collaboration with teams around the world.
The Israeli-led innovations are part of a broader effort by Microsoft to turn AI agents from simple chat interfaces and employee-assistance tools into a fully integrated work environment. In this model, employees and autonomous agents collaborate, manage projects and carry out tasks independently. During his keynote, Microsoft Chairman and CEO Satya Nadella described Copilot as the “new operating system for work.” The platform brings together AI-powered conversation, collaboration and coding, while allowing autonomous agents to work alongside familiar Office applications.
1 View gallery
מטה מיקרוסופט ישראל הרצליה
מטה מיקרוסופט ישראל הרצליה
Microsoft HQ in Israel
(ShU studio / Shutterstock)
One of the key technologies unveiled at the conference was Microsoft IQ, which expands Copilot’s capabilities by giving AI agents access to a broader understanding of an organization. The technology brings together data from Microsoft 365, Fabric, Foundry and third-party systems, giving agents context about an organization’s employees, documents, communications, meetings and projects. Hundreds of new connectors will allow agents to retrieve information from the systems where work actually takes place. Organizations will also be able to consolidate business context into add-ons and share it across teams, while retaining control over their internal knowledge.
Another Israeli-developed technology drawing attention at the conference was Project Perception, an AI agent-based cybersecurity system that Microsoft said is now generally available. It operates through three groups of agents: a “Red Team” that simulates cyberattacks, a “Blue Team” that detects them, and a “Green Team” that works to address the vulnerabilities they expose. The teams operate according to predefined scenarios, while organizations retain control over their actions. Microsoft also introduced its “Cyber 2 Flash” security model and announced that it would be integrated into Project Perception.
The announcements were part of Nadella’s broader vision for bringing AI agents into everyday work. Microsoft introduced Copilot Workspaces, an environment where employees and agents can collaborate on projects using shared information and context. It also unveiled Autopilot, an agent designed to handle repetitive tasks and keep projects moving even when employees are not actively working, without requiring human input at every stage. Copilot Code, meanwhile, enables users to create applications, dashboards and tracking tools through natural-language instructions, with enterprise-level security and controls. Microsoft also announced a partnership with Salesforce to integrate the company's customer relationship management (CRM) system into Copilot.
Nadella outlined three principles guiding Microsoft's development of the next generation of AI tools: putting human aspirations at the center; ensuring organizations retain control over their own knowledge and insights; and building trust by allowing users to oversee agents’ actions and define the boundaries of their authority. The central challenge, he said, is turning rapid technological progress into tangible results for individuals and organizations while preparing for the implications of systems that increasingly act on users’ behalf.
Beyond Microsoft IQ and Project Perception, the company unveiled several other technologies developed with contributions from its Israeli R&D center. In the area of sovereign cloud, Microsoft announced expanded capabilities that allow organizations to run AI models and agents on their own infrastructure, including in environments without an internet connection. Foundry Local is now generally available on Azure Local, allowing models to run in local environments while organizations retain control over their data. Other new capabilities allow agents to access local data, conduct searches and use tools even in environments completely disconnected from external networks.
In cybersecurity, Microsoft announced new capabilities for Defender and Sentinel developed in Israel in collaboration with teams around the world. The updates are designed to accelerate threat detection and response, protect vulnerable systems before security updates are installed, and extend security coverage to AI agents, including those running on third-party platforms. The technology can identify and block malicious actions before agents carry them out.
Another Israeli-developed innovation is Guest RDMA, a technology designed to accelerate data transfers between systems within Microsoft's Azure cloud. It is intended, among other applications, to improve the performance of AI workloads and other data-intensive systems. Microsoft developed the technology in collaboration with Israeli company Silk. In tests conducted by the company, the technology reduced data-transfer latency by as much as 78% compared with standard network communication over the same infrastructure.
Microsoft also announced its AI Transparency Initiative, which aims to promote open standards for transparency, security and governance in AI systems. The initiative is intended to help organizations review the information accessed by AI agents, verify their actions and manage their permissions. As part of the effort, Microsoft will expand its collaboration with the U.S. National Institute of Standards and Technology (NIST). The company said 40 organizations are already participating in efforts to develop and implement the standards.
More than 200 representatives from Israel are attending this year's Ignite conference, including delegations from approximately 70 Israeli startups. Participating companies include Check Point, Varonis, Eon, ControlUp, VAST Data, Cyera, Gong, Semperis and BlinkOps. They are expected to present their technologies to an international audience and meet with senior Microsoft executives, representatives of global corporations, and potential customers and partners. Some are also expected to announce new collaborations with Microsoft during the conference.