Agentic Messaging: The Impact on Customer Interaction & How to Prepare

August 2025
Telecoms & Connectivity

Juniper Research believes that agentic AI will have a substantial disruption on the customer interaction market; not just through improving the capabilities of communications platforms, but also changing how stakeholders generate revenue from customer interaction traffic.

However, before delving into what agentic messaging is, we would like to clarify three distinct terms:

  • AI agents are systems that leverage agentic AI to perform tasks autonomously or semi-autonomously for enterprises, based on user requests.
  • Agentic AI is a form of artificial intelligence. It is a goal-oriented technology that is capable of adaptive decision-making. Unlike previous iterations of AI, agentic AI can alter the environment in which it is situated, such as a business system or network,  based on the goals it has been set.
  • Agentic messaging applies the principles of agentic AI to communication platforms. It involves AI agents that manage conversations across time and context, handling multi-turn dialogues, recalling user preferences, and working across services. Most importantly, agentic messaging can refer to agent-to-agent communications in which agents across various use cases can communicate based on a common goal.

Whilst there is substantial overlap between these terms and what they encompass, there are key differences in their remit, their capabilities, and who uses them. Nonetheless, many of the concepts raised in this blog apply to all three.

Why Agentic Messaging is the Next Step for AI Agents

Agentic messaging is a key opportunity for customer interaction platforms; however, the varying degrees of autonomy that these systems can be given and the impacts they can have on enterprise operations remain the most significant risk to the deployment of agentic AI solutions.

Omnichannel communications, a term that has gained popularity over the last decade, now faces an unprecedented focus on the autonomy of customer interaction. Strategies have primarily focused on reducing the siloes between communication channels and platforms. We believe that the rise of agentic AI will allow enterprises and communication platforms to shift tremendous amounts of processes to agentic AI, which can determine the best outcome for each use case. For example, two similar customer enquiries may require different responses based on various other factors such as customer history or the time of day the request was made. A well-trained agentic AI model will enable it to make optimal decisions in real-time.

The potential impact on customer interactions will be profound as agentic messaging will provide a unified, context-aware journey where agents share information and collaborate to resolve customer enquiries efficiently. Overall, this will lower an enterprise’s operational cost whilst providing scalability in customer interaction.

How Does the Rise of Agentic Messaging Impact Mobile Messaging?

As this market grows, the value chain will become increasingly fragmented as AI specialists are introduced. These AI specialists then claim a portion of the value previously distributed amongst the incumbent players. This may force established players to either increase their prices or reduce their profit margins when introducing AI agents into their technology stack.

Conversational messaging is not a new concept in the customer interaction market; however, it is a concept that has failed to deliver a widely used and scalable monetisation model. We attribute this failure to efforts in trying to implement conversational models into channels primarily designed for one-way communication.
Agentic AI itself will have substantial applications outside of customer interaction; as a result, monetisation models that are not historically based on messaging are required for the technology to be successful. We expect providers to implement a per-token pricing model, which offers a scalable and transparent way to charge for usage based on the actual computational effort required.

For clarity, a token is typically defined as the smallest unit of text or information inserted into either an input to the AI model or an output from the AI model. Whilst there is some variation, an input of “Why is my phone bill higher than usual this month?” would count as 10 tokens given the length of the sentence. Therefore, an AI agent will charge the per-token price for 10 tokens to the end user, who, in the future, is likely to be mobile messaging vendors and CPaaS players, who in turn charge the enterprise using the AI agent.

This model stands in contrast to traditional price-per-message frameworks used in telecom channels, such as SMS and RCS Business Messaging, where enterprises are charged a fixed rate per message sent or received. While effective for static, one-way communication, price-per-message models are ill-suited for agentic AI, where interactions are dynamic, multi-turn between user and enterprise, and context-rich. A single agentic exchange might involve hundreds or even thousands of tokens as the AI reasons, plans, and communicates across systems.

However, agentic messaging extends far beyond pure customer interaction; these AI agents will also require tokens to interact with business systems and other AI agents. As a result, the pricing not only needs to reflect the pricing for conversational messaging over channels, but also the cost of tokens required for agentic AI to make changes to its environment based on user inputs.

What do Mobile Messaging Vendors Need to Do?

Juniper Research has identified several key strategies that mobile messaging vendors should implement to maximise their success in AI agents. Most importantly, these players must leverage their extensive API libraries to deliver this capability to the enterprise end-user. However, charging on a per-API basis and a per-token basis may inflate costs for the enterprise and dissuade initial use.

To combat this, we suggest offering a ‘freemium’ model in which the first agreed-upon number of tokens is sold at a set price, with additional charges then applied for overage. This will enable enterprises to allocate their resources for customer interaction better.

Secondly, these vendors must provide support for agent-to-agent communications. For example, enabling an AI agent based in a billing system to successfully communicate with an AI agent based in a subscription management system. This requires robust protocols and infrastructure that allow agents to interact intelligently and securely.

Conclusion

Agentic messaging will drive the next wave of innovation in customer interaction services through the increased use of innovation. Whilst generative AI has revolutionised the generation of content and understanding, agentic AI provides enterprises with a further benefit: service automation.

However, while Juniper Research believes that agentic messaging will have a substantial impact on customer interaction, the effect of these solutions will grow slowly over time. We recommend that any enterprise wishing to implement agentic messaging into its operators do so gradually, incrementally increasing the amount of autonomy and decreasing the level of human intervention required over time across various systems.


As VP of Telecoms Market Research at Juniper Research, Sam produces high-quality research on telecommunications technologies and the future of digital content. His recent reports include CPaaS, Direct-to-Cell, and 5G Future Strategies. Sam has been interviewed by leading media outlets, including the BBC and Wall Street Journal, and is a regular contributor to messaging conferences and telecommunications industry events.

Latest research, whitepapers & press releases