AI agents are rapidly becoming a core part of modern software systems.
But as organisations connect agents to tools, data, and APIs, a new and largely overlooked security risk is emerging.
Model Context Protocol (MCP) is fast-emerging as the standard way AI agents interact with external systems. Yet most organisations are deploying MCP servers without the security frameworks needed to protect them.
Recent analysis suggests that 92% of MCP servers may be vulnerable to attack, creating a growing attack surface across enterprise environments.
In this live session, two leading experts will break down what this means for organisations building AI systems, and how to secure MCP infrastructure before vulnerabilities become incidents.
What you’ll learn:
- The agent gateway in action: A live demo of how an agent gateway centralizes management and mitigates risk.
- The five critical risks: An autopsy of the most common MCP vulnerabilities.
- Cost & instrumentation: Strategies to gain visibility into agent behavior.
- A practical governance model: Create controls without slowing innovation.
Speakers
Paul Turner
VP Market Strategy, Tray.ai
Paul is responsible for the AI Orchestration product strategy at Tray.ai. He has over 30 years of experience in enterprise software, applications, and middleware in product management, strategy, and consulting across ERP, HCM, data warehousing, application integration, and analytics, in senior roles at Workday, NetSuite, and other SAP and Oracle companies.
Luke Smith
Principal Sales Engineer, Tray.ai
Luke Smith is a Principal Sales Engineer at Tray.ai, where he has spent four years helping customers turn AI and agents into real business performance. Luke specialises in Tray's Merlin Agent Builder, Agent Gateway, and MCP tooling, working closely with customers to build intelligent agents, orchestrate AI across their enterprise stack, and scale automation with built-in governance.
