Model Context Protocol (MCP), AI & the Increased Use of Natural Language to Interact with CSPs’ Systems
This article was first published in The Fast Mode.
As Agentic AI continues to roll out across many functions in Communications Service Providers (CSPs), the Model Context Protocol (MCP) is poised to become a major topic in 2026. Introduced by Anthropic in November 2024, MCP is an open-source layer that standardises how AI systems integrate and share data with external tools, systems, and data sources.
Beyond Anthropic, MCP is being widely adopted by companies such as OpenAI and Google, as well as wide ISV industry to enhance their AI systems, and it is rapidly becoming the standard for AI integration. Anthropic has donated MCP to the Linux Foundation’s Agentic AI Foundation (AAIF) to promote its development as a neutral, open, community-driven standard. This strengthens the position of MCP as an industry standard rather than a proprietary technology. https://www.anthropic.com/news/donating-the-model-context-protocol-and-establishing-of-the-agentic-ai-foundation
In the telco software landscape, MCP can be used as an API layer to enable and power AI-driven experiences in both end-user and business user tools as it standardises bidirectional interactions between models and systems. It provides a layer, for example, in BSS solutions that integrates with existing APIs such as TM Forum’s Open APIs. This offers a consistent and standardised way for AI-native apps and experiences to access needed information and services.
For CSPs, who operate in extremely complex, siloed, and heavily regulated environments, MCP solves several long-standing structural challenges. It defines a structured way for AI models to access and use data, tools, and user context efficiently. Essentially, MCP acts as a universal “plug-and-play” interface between AI systems (like OpenAI’s GPT and Anthropic’s Claude) and business resources such as CRM information for users’ services, offer catalogues for available services and experiences, knowledge bases and documentation systems, APIs, automation tools, and various workflows.
Increased use of Natural Language to Interact with BSS and Monetization Systems
Instead of each AI integration being custom-coded for every app, MCP provides a shared standard, so MCP-compliant systems can interact with each other. This means that CSP users (such as customer care or sales representatives) can use natural language to interact with systems such as BSS and monetization suites.
MCP servers provide an interoperability layer that enables different AI experiences to integrate with CSPs’ systems in their business operations, service operations, and resource operations. This opens up a wide range of operations in a CSP to be supported with AI. To illustrate the importance of MCP for CSPs, consider the following examples of how MCP can transform day-to-day operations in B2B sales, customer care, B2C sales, marketing, and personalisation.
Agentic AI Integration Through MCP – Sample Impacts on Operational Areas
B2B Sales
MCP enables AI agents to access CRM data for customer profiles and history, sales data for existing cases, and network management systems, including digital twins. For example, a user might ask, “Can you create an offer for this location on this date to deliver guaranteed connectivity to support 100 hi-def video applications and 300 EPOS terminals?” The AI agent can create the offer, validate its performance against the autonomous network’s digital twin, and suggest selling a service-level agreement (SLA) to the customer.
Customer Care
AI agents can access CRM and BSS for customer profiles and history, ticketing systems for open issues, network monitoring for service status, and billing platforms for payment data. When a customer asks, “Why is my internet slow?” the customer care rep, assisted and prompted by an AI agent, can respond with a personalised, contextual explanation, suggest an upgrade, or schedule a technician visit—without exposing private data. This leads to faster care call resolution, higher net promoter score (NPS), and reduced workload for customer care teams.
B2C Sales, Marketing, and Product Management
MCP allows AI to access billing data (spend, top-ups, payment delays), usage analytics (data consumption, roaming patterns), CRM (customer lifetime value, churn probability), and product management (rapid new offer creation). For example, AI can be asked to provide current retention offers for declining data usage and medium churn risk, and further to create a new offer variant for updated characteristics and marketing proposition.
As we see the rise of AI in CSPs, we will see increased use of MCP servers to integrate AI tools with CSPs’ systems. This will drive a change in how users interact with systems such as BSS and monetization stacks, as an increasing number of users use AI tools and natural language as the primary way to interact with CSPs’ systems.
Topias Koskimäki
Head of Product Management