MCP: Can AI Agents Make Customer Interactions Better?

July 2025

Telecoms & Connectivity

Molly Gatford

Senior Research Analyst

This year, we have seen a range of global companies adopt the Model Context Protocol (MCP). Just this month, Sinch has also announced that it has implemented MCP for its AI agents. With so many large companies implementing MCP, this is a significant step in the rise of AI agents for automating tasks within customer interactions. But what is MCP, and why is it important?

MCP: Allowing AI Agents to Interact with Data Sources in Real-time

MCP was introduced by Anthropic in November 2024, and the standard solved a major challenge that arises when connecting AI agents to external data sources and tools; it replaces fragmented, proprietary integrations with a single, universal protocol.

Simply, MCP is an open-source standard that enables AI agents (MCP clients) - which are often orchestrated by large language models (LLMs) - to interact with the servers that expose databases, tools, and external systems. 

Platforms that provide AI agents for use in customer interactions, such as Gupshup, Infobip, and Sinch, have all implemented MCP. This enables more seamless integration with servers that have also implemented MCP. Notably, Salesforce has introduced MCP-hosted servers, Shopify offers the Shopify.dev storefront MCP server to connect to Shopify store's catalogue, shopping cart, and policies, and Stripe also offers an MCP server for AI agents to interact with the Stripe Application Programming Interface (API).

The below figure illustrates how this system operates in practice, in a scenario where a customer requests information on stock with the intention of placing an order and completing payment. It shows how an MCP client might interact with multiple servers and, therefore, how MCP - as a standardised protocol - makes it much easier for connecting with each server.

If proprietary integrations were required each time a business wanted to connect an MCP client to an MCP server, it would take far longer and be much more costly to launch. As a result, Juniper Research anticipates that the implementation of MCP will accelerate the adoption of AI agents. 

Customer Interaction with Brand Leveraging AI Agents (MCP Client) & MCP Server

Source: Juniper Research

Impact on AI Agent Adoption

We believe that future growth in business messaging traffic will be attributed to the increased adoption of conversational use cases by large enterprises. With the expectation that an increasing number of customers will prefer messaging a brand rather than calling, enterprises are expected to adopt AI agents to automate a wider range of customer interactions; with conversational commerce, and customer support, being key use cases.

The cost of conversational AI solutions is decreasing for enterprises, with factors such as model compression and the creation of smaller, lightweight models being key. This is resulting in a declining cost with improving technology, which will grow adoption of these solutions; especially with AI agents that can fully automate a greater number of customer interactions.

The growth in the number of chatbot interactions, as shown below, will, in part, be attributed to the increased adoption of AI agent solutions by brands. As a result, we believe that AI agents will improve customer sentiment towards chatbots. With increased adoption by brands, there is a greater chance of customers interacting with chatbots in their daily lives. 

Total Chatbot Interactions (m), 2025-2029, Split by Channel

Source: Juniper Research

Juniper Research predicts that web-based and app-based chatbot interactions will continue to account for a significant proportion of all chatbot interactions over the next five years. Web-based chatbots have the benefit of being easily accessible to customers, whereas app-based chatbots are important for industries such as banking and healthcare, where a customer typically goes through multi-factor authentication to sign-in to the app, which ensures added security for app-based chatbots handling customer data.

OTT messaging chatbots are expected to experience significant growth over the next five years. One of the reasons for this will be the growth in adoption of WhatsApp for conversational business messaging use cases. This is as WhatsApp has made it free to respond to customers in certain templates; making it more attractive for enterprises to use. Also, the click-to-chat ad feature will be useful for brands to acquire new customers and engage with existing customers, through the placement of these ads in Meta apps. RCS business messaging is not as well-suited to conversational commerce interactions currently, with the entry points being less clear for customer-initiated interactions.

Communications platforms will benefit from the growing demand from enterprises for conversational use cases; as these use cases offer higher margins compared to transactional use cases such as authentication, notifications, and reminders, given the requirement for multiple SaaS components. Therefore, Juniper Research anticipates a growth in the number of communications platforms seeking to offer AI agents, with a priority to ensure these utilise MCP so that they can connect to external databases.

To conclude, MCP is a crucial open-source protocol that is expected to accelerate the adoption of AI agents; thereby expanding the number of tasks that can be automated as part of the end-to-end customer journey.

However, this is not the only factor at play; it will be essential for communications platforms to provide detailed analytics on the performance of AI agents and the number of customer requests they can automate. Doing so will be essential to demonstrating the return on investment of these solutions for brands, as it will only make sense to implement AI agents for use cases where they can save a significant amount of time for human agents, thereby decreasing costs associated with customer interaction. 

Molly is a Senior Research Analyst at Juniper Research, providing insight, data, and recommendations for established and future markets within the telecommunications sector. Her recent reports have covered topics including CPaaS, Conversational AI, and RCS Business Messaging.