Representative End User Clients
Representative Automation Clients
Representative Software Clients
As many of you know from my recent three-part blog series, I’ve been enthusiastically exploring the transformative potential of combining MCP, UNS, and CESMII to shape the future of Industrial AI:
Context Is the Missing Link: The Emergence of the Model Context Protocol in Industrial AI
Contextual Harmony: Industrial AI, Unified Namespaces and the Promise of MCP
The Power Trio: MCP, UNS, and CESMII–Orchestrating the Future of Industrial AI
The vision of a unified, contextualized data landscape enabling scalable AI deployments across manufacturing is incredibly compelling. I wanted to highlight MCP’s broad applicability, and indeed, there's growing excitement within the industry about its potential.
However, as we at ARC Advisory Group well understand—especially in the complex and often heterogeneous world of manufacturing—the devil is always in the details. While that series focused on MCP’s immense promise, it wasn’t meant to gloss over the practical realities. Instead, it was a call to our community—a rallying cry to explore and engage with MCP more deeply, so we can collectively unlock its true potential and pave the way for widespread Industrial AI adoption.
Since then, I’ve had the opportunity to connect with several real-time systems architects, whose thoughtful feedback has surfaced key limitations in the current MCP framework—particularly around security, identity, and data transport. I’m grateful for their insights, which underscore the importance of constructive dialogue. This follow-up aims to address some of these challenges and encourage broader engagement across our Industrial DataOps community.
Navigating the Nuances: Security and Identity in MCP
There’s no question about the excitement around MCP’s ability to deliver rich, contextualized data. Yet, in discussions with systems experts, several important security and identity management considerations have emerged:
Potential for Contextual Data Leaks: MCP’s strength—persistent context—can also be a vulnerability. Without strong security controls, long-lived context sessions could inadvertently expose sensitive operational data. Managing the data lifecycle and enforcing robust access controls is critical.
Metadata Exposure Risks: The protocols underpinning MCP often carry metadata related to sessions and interactions. Understanding what’s exposed and protecting this metadata is key to maintaining operational confidentiality.
Robust Identity Management Is Essential: As MCP adoption grows, well-defined identity and access management becomes non-negotiable. Restricting access to authorized systems and users is fundamental to ensuring secure deployments.
These aren’t insurmountable issues. But they are crucial areas where the Industrial DataOps community must contribute as MCP evolves. Secure, trustworthy systems demand active engagement and thoughtful solutions.
Embracing the Evolution: From HTTP+SSE to Streamable HTTP
Another significant development in MCP is the shift from HTTP+Server-Sent Events (SSE) to “Streamable HTTP.” While this might seem like a technical nuance, it signals a meaningful evolution in how real-time data could be exchanged within the MCP ecosystem.
HTTP+SSE offered a lightweight, server-to-client streaming model. “Streamable HTTP,” however, appears to introduce a more flexible, efficient method of real-time communication—one with several potential benefits:
Improved Real-time Performance: Especially valuable in industrial environments where high-frequency data is the norm, streamable protocols may offer reduced latency and improved responsiveness.
More Flexible Communication Patterns: Moving beyond SSE’s unidirectional model could unlock more sophisticated and responsive interactions between systems.
Alignment with Modern Web Standards: Adopting familiar HTTP streaming patterns can improve interoperability and simplify integration across existing infrastructures.
This shift deserves attention—not just from developers, but from all of us interested in building real-time, AI-ready industrial systems.
A Call to Action for Industrial DataOps and AIOps Professionals
As Director of Research for Industrial AI at ARC Advisory Group, I believe the nuances around security, identity, and transport mechanisms are precisely why our community must take an active role in shaping MCP’s future. Here’s where your expertise can make a real difference:
Shape Security Best Practices: Your insights into OT/IT security are essential in identifying and mitigating risks in MCP, helping guide the development of secure implementation guidelines.
Drive Real-time Data Innovation: Understanding how “Streamable HTTP” works in practice will allow you to create more efficient, scalable real-time data pipelines—key infrastructure for Industrial AI.
Champion Interoperability: Your practical experience with diverse industrial systems is critical to ensuring MCP integrates seamlessly across environments.
Accelerate Scalable AI: By engaging with and refining MCP, you’re contributing directly to our shared goal: enabling Industrial AI to scale meaningfully and reliably.
My initial excitement about the potential of MCP, UNS, and CESMII hasn’t dimmed. But real progress demands both optimism and critical thinking. Let’s take the feedback from our peers seriously, and as a committed Industrial DataOps community, continue shaping MCP into the backbone of the Industrial AI R(E)volution.
Engage with ARC Advisory Group
For ARC Advisory Group recommendations for Navigating the AI Wars, Closing the Digital Divide by Embracing Industrial AI, assembling your Industrial-Grade Data Fabric, and governing and guiding major decisions about enterprise, cloud, industrial edge, and AI software, please contact Colin Masson at [email protected] or set up a meeting with me, or my fellow Analysts at ARC Advisory Group.