Will Agentic AI kill customer service channels?

Customer-centric businesses have matured effective and efficient omnichannel customer experience strategies. A growing body of evidence points to their value. Research has found that 45% of firms leverage better customer engagement through omnichannel experiences.  Customers are nearly four times (3.6x) more likely to make additional purchases. And omnichannel marketing has proven its potential to increase sales by 287%. 

Meanwhile, research has also found that 80% of customers think their experiences with brands should be better considering all the data that companies collect. In an age when consumers value their time more than ever before, we are entering a new phase of AI-powered experience. Customers will no longer want to ‘contact’ a company via phone, email or chat. Instead, they will simply ask their personal AI assistant to solve their problem for them. 

Imagine this scenario in the very near future: you are involved in a minor car accident. Instead of searching for the towing company’s number, filling out an accident report, or calling your insurance broker, you simply say, "Hey, AI. I had a minor fender bender. Handle roadside assistance and open a claim with my insurer. Send them the photos I just took."

This is the paradigm shift. Customers will delegate problem-solving to their ubiquitous AI assistants that are integrated into their devices, cars and homes. Each assistant acts as the sole intermediary between their user and the brands. For the customer, the experience will be seamless, effortless and, most importantly, ‘channel-less’. They won't care how their assistant solves the problem. Only that it does.

So, if customers no longer interact with their service providers directly, how do these companies communicate with them? The answer is through an ecosystem of Application Programming Interfaces (APIs). However, for this ecosystem to function at a global scale and avoid becoming a chaotic assortment of different proprietary connections, open standards are essential. This is where emerging protocols come into play—protocols that will define the future of communication.

Two of the most promising of these are the Agent-to-Agent (A2A) and Model Context Protocol (MCP) protocols. These are not competitors but rather complementary.

• A2A: the language of collaboration. Think of A2A as the protocol that defines what AIs are discussing. It's a shared business language that enables the customer's AI assistant and the company's AI agent to discover each other, understand each other’s capabilities, and negotiate complex tasks. It standardizes requests like "open a claim", "request documentation", or "confirm an appointment with the appraiser". This layer ensures that both AIs speak the same functional language.
• MCP: the means of communication. This protocol operates at a lower level, establishing a secure and standardized channel for an AI to access the tools and data it needs. It's the ‘USB-C of AI’, enabling the insurance agent to securely connect to its policy database or to an external appraisal system to get the necessary information for the A2A conversation.

In this new model, the connection between the customer's personal assistant and the company's agent will use A2A so that both parties can collaborate and solve the problem ‘automatically’. At the same time, enterprise systems (modern and legacy) are connected via MCP within the company's agentic AI. The ‘conversation’ is a structured transaction between two systems that collaborate autonomously; the human agent only intervenes when human judgment or decision approval is required.

For decisions to be approved and binding transactions implemented, a foundation of absolute trust is required. Traditional username and password systems are completely insufficient for this new paradigm. Security must evolve and adopt new standards for decentralized digital identity.

The key concept is that control of identity returns to the user. Each customer will manage their data and permissions through a personal ‘wallet’. This wallet will contain ‘verifiable credentials’ (such as their national ID, insurance policy or driver's license) issued by trusted entities such as governments or corporations. This establishes a robust digital governance framework based on three pillars:

• Identification: an initial process will exist to reliably generate these verifiable credentials, just as we currently obtain our national ID or passport.
• Authentication: the customer's assistant cryptographically (that is, using cryptography) proves its authenticity and has the user's permission to act. It does so by presenting proof from its wallet, without needing to share passwords or sensitive data that could be stolen.
• Authorization: the wallet itself manages the granular permissions the user has granted to their assistant. The company can cryptographically verify if the agent is authorized to view data, approve a charge up to a certain limit, or modify policy details.

Without this new approach based on user-sovereign identity, the system would collapse as a result of fraud and distrust. This has already started to happen given the increase in spam across all channels, where customers cannot necessarily trust who is contacting them via phone, SMS, emails or WhatsApp.

Agentic AI is already reshaping business. Gartner predicts that by 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024. And in the same timeframe, at least 15% of day-to-day work decisions will be made autonomously through agentic AI, up from 0% in 2024. 

Given this trajectory, priorities for forward-thinking companies are based on three strategic objectives. First, simplify the customer's life by unifying their experience via a single point of contact: their personal assistant. Second, connect these assistants with company ‘agentic’ systems through APIs and open standards that ensure interoperability. And third, implement security and traceability mechanisms based on decentralized identity that provide guarantees to both parties in every automated transaction.

In short, the next frontier of customer experience is already defined. The future is here—the channel-less experience revolution has begun.