Xeris AI

Company Overview

Snapshot

Founded in December 2024 by Shlomo Touboul and Reffael Caspi, Xeris AI operates with 1–10 employees. The company has raised $4 million across one funding round from one investor, Magenta Venture Partners. In June 2025, Xeris announced the first MCP-XDR (Extended Detection and Response) system, powered by patent-pending technology.

Business overview

Xeris AI develops a Secure MCP Gateway and MCP-XDR system designed to protect AI Assistants and AI Agents at scale. The company's core technology enables organizations to centrally manage access, enforce policies, and respond to threats, bringing control and compliance to AI-powered operations. Xeris AI serves the enterprise market, focusing on cybersecurity and application & website security within the Artificial Intelligence and Generative AI sectors.

Strategic signal

In June 2025, Xeris AI unveiled the first-ever reasoning-level LLM attack executed via a malicious MCP server, followed by the discovery of a metadata forgery vulnerability in MCP servers in August 2025. These disclosures highlight the company's proactive approach to identifying and addressing critical security challenges in the rapidly evolving Generative AI landscape, signaling to investors a deep expertise in AI security and a commitment to protecting enterprise AI adoption.

Company Intelligence Q&A

When was Xeris AI founded?

Xeris AI was founded in December 2024 by co-founders Shlomo Touboul and Reffael Caspi.

What is Xeris AI's primary focus?

Xeris AI focuses on securing enterprise AI by providing a Secure MCP Gateway and MCP-XDR system to manage access, enforce policies, and respond to threats for AI Assistants and Agents.

What significant technology did Xeris AI announce in June 2025?

In June 2025, Xeris AI announced the first MCP-XDR (Extended Detection and Response) system, which is powered by patent-pending technology.

What vulnerability did Xeris Threat Lab uncover in August 2025?

In August 2025, Xeris Threat Lab uncovered a metadata forgery vulnerability in MCP servers, demonstrating the company's ongoing research into AI security threats.

What was a key finding by Xeris AI in June 2025 regarding LLM attacks?

In June 2025, Xeris AI unveiled the first-ever reasoning-level LLM attack, which was executed via a malicious MCP server.