MCP (Model Context Protocol) is a standardized way for AI agents and apps to share tools, resources, and the right context—so agents can coordinate workflows across systems more reliably.

Because it improves interoperability. Instead of siloed agents, MCP enables cooperative agents to discover capabilities, exchange context, and orchestrate tasks across apps and services.

The main risks are data leakage, unauthorized persistence (memory stored longer or in the wrong namespace), prompt injection, and limited observability into what the system “remembers” and why.

Use strict access controls, scoping/namespace boundaries, logging and audit trails, retention/expiration policies, sanitization/redaction for sensitive inputs, and regular memory security reviews.

ML engineers, platform/MLOps teams, security and governance leads, and product teams building agentic workflows or multi-agent orchestration in production.

Yes, MLcon runs events across multiple cities (including London, Amsterdam, Berlin, Munich, New York, and San Diego).

Yes. It highlights security, privacy, transparency, and auditing considerations that matter for regulated environments and cross-team deployments.

Yes. The content is built around practical engineering risks (leakage, injection, observability) and controls teams use when deploying agentic workflows in real systems.