Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

The future of customer experience is AI agents working together

Contact centers are evolving to autonomous systems delivering end-to-end resolutions.

[Photo: Getty Images]

Profile photo of Al Martin

VP Customer Experience · Forethought

A chatbot is still a chatbot. And that’s the problem.

While GenAI has helped contact centers deliver better automated experiences, these gains are strictly incremental, not transformational. Customers still end up doing most of the work.

The real potential lies in teams of AI agents that collaborate across sales, marketing, support, and customer success functions. Teams of AI agents won’t just answer questions. They’ll work together to identify problems, determine solutions, and use specialist AI agents to deliver end-to-end resolutions for complex customer issues.

This isn’t an aspirational long-term vision. It’s what we should expect to see at leading customer experience (CX) organizations in the immediate future.

The benefits of a multi-agent CX system

Agentic AI is projected to handle 68% of all customer service and support interactions by 2028. The question is no longer where to deploy AI, but how to do so in a way that unlocks its full value. By mirroring the way human organizations work—with specialized personnel collaborating to understand problems and taking action to solve them—a multi-agent CX system offers a new way to deliver customer experience at scale.

Human support organizations don’t expect isolated generalists to be able to handle every issue. Neither should multi-agent CX systems.

Instead, multi-agent CX systems should deploy focused AI agents to address different aspects of the customer journey—one for billing, another for technical support, and so on. An orchestration layer routes requests intelligently based on query type, customer context, and agent availability, and enables multiple agents to work simultaneously on complex problems. This modularity makes systems easier to extend and maintain while improving accuracy through specialization.

Interoperability is the real game-changer

Many organizations obsess over which AI model to deploy. What matters most is interoperability—the ability of agents to discover tools, share context, and take safe, auditable actions across the entire technology stack.

Consider the way a multi-agent architecture would assist a customer who calls about a delayed shipment:

  • One agent would retrieve shipping information.
  • Another would check inventory.
  • A third would generate solutions.

Each of these actions requires integration with the appropriate enterprise system, including authentication, credentials, and API access. Without standardization, just a handful of specialized CX agents could cause complexity to skyrocket, creating a massive engineering bottleneck.

Model Context Protocol (MCP), a new open standard that defines how AI systems connect to data and tools, helps solve the interoperability challenge by functioning as a universal adapter for AI agents. MCP creates a standardized, two-way connection for LLMs to request help and perform actions through external resources.

Interoperability unlocks the value of multi-agent CX. By enabling agents to discover tools, share context, and take safe, auditable actions across the tech stack, organizations can avoid friction, delay, and dropped context to deliver a better customer experience.

How organizations can prepare

To manage a multi-agent operating model successfully, CX leaders will need to ensure they have the right people on their teams managing tickets, tech, operations, and knowledge to tie everything together.

Although leaders won’t need engineering skills, they will need a baseline of technical fluency. A working knowledge of orchestration patterns, integration approaches, and data pipelines is essential to understand the tradeoffs involved in architecture decisions. Integration readiness checklists can determine which integrations will be straightforward and where custom development may be needed to help CX teams secure scarce engineering time.

Product management thinking will also help leaders define clear roles and success metrics. Questions to ask include:

  • What are the problems our customers face?
  • What agents should we use to resolve these?
  • What kind of metrics should we use to measure performance: containment percentage, average resolution time, first-contact resolution, customer satisfaction?
  • How can the agent drive continuous improvement?

Workflow design skills will be critical. CX leaders will need to decide which queries agents can handle autonomously and when to escalate a query to humans. This can be a nuanced call; in some cases, agents should persist by asking clarifying questions, rephrasing prompts, or consulting a knowledge base. In other cases, a quick human assist may be key for a positive customer experience. When it’s appropriate to escalate to a human, the handoff must be seamless and efficient, retaining full context so the human agent can continue the conversation as naturally as if they’d been involved from the beginning. 

At a high level, the path to success for multi-agent CX isn’t markedly different from traditional thinking. Organizations that treat agents like a core part of the support organization with clear roles, tool access, and measurable outcomes will turn AI from a pilot into an effective and durable resolution engine across every customer moment.

Al Martin is vice president of customer experience at Forethought.