Today in “Connection Center” news, Avaya became another adopter advocate announcing that its Avaya Infinity Platform will support Model Context Protocol (MCP) later this year. Model Context Protocol, created by Anthropic, is the open standard that allows AI models to securely and reliably interact with external tools, data sources, APIs, and user context.

As part of this initiative, Avaya is collaborating with Databricks, a leading data and analytics provider simplifying and democratizing data and AI, to deliver enterprise-grade data security and governance at scale. It’s good when traditional Contact Center space legends stretch out to bridge in innovation.

Contact center businesses like Avaya can gain several key advantages by incorporating Model Context Protocol (MCP) into their technology stack, and the right strategic partnerships:

Unified AI Integration MCP provides a standardized way to connect AI models with various data sources and tools across the contact center ecosystem. This means Avaya can integrate AI capabilities more seamlessly across their CRM systems, knowledge bases, call routing systems, and agent desktop applications without building custom connectors for each integration.

Enhanced Agent Productivity With MCP, AI assistants can access real-time customer data, interaction history, and knowledge articles simultaneously during customer interactions. Agents get contextually relevant suggestions, automated call summaries, and intelligent routing recommendations without switching between multiple systems.

Improved Customer Experience The protocol enables AI to maintain context across different touchpoints – phone, chat, email, social media – providing consistent, personalized service regardless of channel. AI can access comprehensive customer profiles and interaction history to deliver more accurate and relevant responses.

Scalable AI Architecture MCP’s standardized approach reduces the complexity of deploying AI across different contact center functions. New AI capabilities can be added more easily, and the system can scale without requiring extensive custom development work for each new integration.

Better Analytics and Insights By providing AI models with access to diverse data sources through a common protocol, contact centers can generate more comprehensive analytics about customer sentiment, agent performance, and operational efficiency across all channels and systems.

The protocol essentially acts as a universal translator that allows AI to work more effectively with the complex, multi-vendor technology environments typical in enterprise contact

Strategic partnership with Databricks Partnerships like this essentially transform contact centers from reactive service points into proactive, data-driven customer relationship hubs. It exponentially amplifies the advantages by introducing advanced data intelligence and lakehouse architecture capabilities. Databricks’ unified foundation for data governance, powered by its Data Intelligence Engine, would consolidate all contact center data – call recordings, chat transcripts, CRM data, and performance metrics – into a single, MCP-accessible repository. This integration would enable sophisticated predictive models for customer behavior and sentiment analysis, while Databricks’ generative AI capabilities would allow contact center AI to create more contextually relevant responses and personalized communications without sacrificing data privacy. The combination of MCP’s standardized AI integration with Databricks’ machine learning operations and real-time analytics would transform contact centers from reactive service points into proactive, data-driven customer relationship hubs.

• Contact center managers could leverage conversational AI interfaces to query complex performance data using natural language.
• Agents would benefit from true omnichannel intelligence that analyzes customer interactions across all touchpoints in real-time, providing comprehensive insights about customer intent and optimal next actions.

Architecture

The Integration Challenge and MCP’s Solution

Enterprise organizations typically operate with hundreds of tools and data sources – from Salesforce and product databases to knowledge repositories and cloud services – while deploying multiple AI models and applications that need to access this information. Traditional approaches require unique, point-to-point integrations between each tool and each AI system, creating a combinatorial explosion where connecting N tools to M AI models demands N×M custom integrations. This architecture becomes prohibitively expensive to build, increasingly difficult to maintain, and fundamentally unscalable as organizations grow.

MCP eliminates this complexity by introducing a standardized intermediary layer that transforms the N×M integration problem into a simple N+M solution. Each tool requires only a single MCP server wrapper, while each AI model needs just one MCP client implementation. Once any component can “speak” MCP, it instantly communicates with every other MCP-compliant component in the ecosystem. This architectural shift dramatically reduces development overhead, eliminates redundant integration work, and prevents the accumulation of technical debt from maintaining dozens of fragile, custom connectors. The result is a truly scalable foundation where adding new tools or AI capabilities requires minimal additional integration effort.

MCP Server

Acts as an intelligent middleware layer that seamlessly connects client applications to external services including databases, cloud platforms, containerized services, and productivity tools. Each MCP server abstracts the complexity of underlying APIs, transforming diverse service interfaces into a standardized format. For instance, an AWS-connected MCP server simplifies cloud compute management, deployment orchestration, and file storage operations into unified commands.

MCP Local Server: Extends capabilities to local environments, enabling direct access and processing of local file systems including documents, images, spreadsheets, and other data formats without external dependencies.

Client Host (Application)

Serves as the primary interface layer – whether frontend application or core processing engine – that users and other services interact with directly. Rather than managing multiple API connections, the client host routes all external service requests through MCP servers via the standardized MCP Protocol. This architecture enables seamless communication with diverse services while maintaining clean separation between user interfaces and backend service complexity.

Model (AI Agent)

Functions as the intelligent decision-making engine that receives contextual information and available tool inventories from MCP servers through the standardized protocol. The model (such as Claude, GPT, or other LLMs) analyzes user requests, determines the appropriate sequence of actions, and orchestrates tool usage across multiple services. Rather than being limited to pre-programmed responses, the model dynamically selects and combines available tools to accomplish complex, multi-step tasks while maintaining awareness of context, constraints, and user intent throughout the interaction.

MCP Protocol

An open-source, standardized communication framework designed specifically for AI agents and LLM workflows. The protocol ensures consistent data exchange formats and interaction logic across all service integrations, regardless of the underlying technology stack. By providing a unified approach to tool-calling and service communication, MCP dramatically reduces integration overhead for chatbots, data pipelines, and enterprise AI platforms, allowing developers to focus on core business logic rather than connection management.

Bold. Smart move!