How MCP Became the Standard for Agentic AI Integration

Key Takeaways

Model Context Protocol (MCP) has become the universal open standard for connecting AI agents to enterprise tools and data, backed by every major technology company and governed by the Linux Foundation. Avaya Infinity^® was built on this architecture before the market converged. Customers on Avaya's platform inherit a compounding advantage as the MCP ecosystem grows.

• MCP reached 97 million monthly SDK downloads in March 2026, up from roughly 2 million at launch. Kubernetes, now considered foundational cloud infrastructure, took close to four years to reach a comparable level of adoption. 
• Every major AI provider, including Anthropic, OpenAI, Google, Microsoft, and AWS, now ships native MCP support. 
• More than 10,000 public MCP servers exist today, up from roughly 1,200 in early 2025, an increase of more than 8x in one year. 
• Gartner predicts 40% of enterprise applications will include task-specific AI agents by the end of 2026.

Enterprise AI is moving from experimentation to execution. The question is no longer whether AI agents will become part of the enterprise technology stack. The question is how those agents will securely connect to the tools, data, systems, and workflows that enable action.

That is why the rise of Model Context Protocol, or MCP, matters.

Introduced by Anthropic in November 2024, MCP was created as an open standard for connecting AI assistants to the systems where enterprise data lives, including business tools, content repositories, development environments, and operational systems. In practical terms, MCP provides AI applications with a more consistent way to access context and interact with approved tools, rather than relying on disconnected, vendor-specific integrations.

That may sound technical. But for enterprise leaders, the implication is strategic.

AI agents are only as useful as the systems they can reach, the context they can understand, and the actions they can take safely. Without a shared integration standard, every model, application, data source, and workflow risks becoming another point-to-point connection. That creates the same fragmentation enterprises have spent years trying to overcome.

MCP changes the architecture conversation. It gives the market a path toward more open, interoperable AI agent integration. It gives enterprises a way to avoid being locked into a single vendor’s model, tooling, or orchestration layer.

The Market is Moving Toward Open Standards

MCP’s momentum did not come from a single vendor. It reflects a broader shift across the enterprise software and AI ecosystem.

OpenAI supports MCP through its agent development ecosystem, including MCP support in the Responses API and Agents SDK. Salesforce has positioned MCP as part of Agentforce interoperability, helping agents connect across systems and access siloed data. Microsoft and GitHub have also publicly supported MCP and joined the broader ecosystem, advancing the standard.

The signal is clear: major AI and enterprise software platforms are aligning around MCP because agentic AI needs a secure, scalable, and interoperable way to connect with enterprise systems.

That matters because the AI agent landscape will not stay fixed. New models, tools, assistants, and enterprise workflows will continue to emerge. A closed architecture forces organizations to follow one vendor’s roadmap. An open standards-based architecture gives enterprises more flexibility to adapt as the market changes.

Governance Changed the Risk Equation

One of the most important milestones came when Anthropic donated MCP to the Agentic AI Foundation, a directed fund under the Linux Foundation, with founding contributions from Anthropic, Block, and OpenAI.

That move matters because enterprise adoption depends on trust. Before MCP adopted neutral governance, some organizations might have viewed it as one vendor’s protocol. Linux Foundation stewardship changes that conversation. It gives MCP a more durable governance model and reinforces its role as shared infrastructure for the next phase of agentic AI.

For enterprises, that lowers the perceived risk of building around MCP. It also strengthens the case for open orchestration: AI systems should be able to connect across approved tools, models, and data sources without forcing the business into a closed ecosystem.

Why This Matters for Customer Experience

Customer experience is one of the clearest places where this architecture shift matters.

Contact centers and CX environments are already complex. They depend on CRM systems, knowledge bases, case management, workforce tools, analytics platforms, communication channels, routing logic, compliance requirements, and customer history. Adding AI agents without an open integration strategy can worsen that complexity.

An AI agent may need to understand customer context, retrieve account information, summarize prior interactions, recommend next steps, trigger a workflow, or hand off to a human employee. In a closed architecture, those actions depend heavily on the limitations and permissions within a single vendor ecosystem. In an open architecture, organizations have more flexibility to connect the right AI capabilities to the right enterprise systems with stronger control over how data and workflows are used.

This is becoming more urgent as AI agents move into enterprise applications. Gartner predicts that 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from less than 5% in 2025. That shift will put more pressure on organizations to decide how agents should connect, what they can access, how they are governed, and how quickly the business can adapt as new capabilities emerge.

Avaya’s Bet Was on Openness, Not Lock-in

This is the larger point behind Avaya’s MCP direction.

Avaya’s decision to support MCP reflects an architectural choice, not just a product feature. It supports the belief that enterprises should have the flexibility to choose their AI models, connect their own systems, preserve control over their data, and evolve their CX environments without being constrained by a single-vendor approach.

Avaya has announced MCP support for Avaya Infinity^® as part of its broader open architecture strategy. Avaya has also announced a partnership with Databricks to support enterprise-grade governance and security for AI at scale. Together, these moves point to the same principle: AI in the enterprise has to be open enough to adapt, secure enough to trust, and governed enough to operate in high-stakes environments.

For customer experience leaders, that distinction matters. The future of AI-powered CX will not be defined by who adds the most AI features the fastest. It will be defined by who can orchestrate AI, data, workflows, and human engagement in a way that remains flexible, secure, and enterprise-ready.

Open Standards Are the Foundation for Agentic AI

MCP’s rise shows that the market is moving toward a shared answer for AI agent integration. The standard is still evolving, and enterprises still need strong governance, security, data controls, and implementation discipline. MCP does not eliminate the complexity of enterprise integration.

But it does create a more open foundation for what comes next.

That is why the open standards conversation matters. As AI agents become more embedded in enterprise applications, organizations will need architectures that can keep pace with new models, new tools, and new customer expectations. Closed systems may offer short-term convenience, but they can limit flexibility over time.

The advantage of MCP is not simply that it connects AI to tools. It helps enterprises avoid rebuilding the same integration problem again inside the AI era.

For Avaya, the bet on open standards was a bet on customer choice, enterprise control, and long-term adaptability. As MCP becomes a more central part of the agentic AI ecosystem, that decision looks less like a technical preference and more like the direction enterprise AI is heading.