Why communication is enterprise AI’s next bottleneck

It’s clear that many organizations are getting AI right — and seeing the benefits. But should they have solved another problem first?

Over the past two years, there’s been a focus on AI model performance, including accuracy, latency, reasoning and multimodality. That made sense for a while, but as AI moves from isolated copilots to interconnected, agent-driven ecosystems, a new constraint is emerging: coordinated AI communications. 

The next wave of enterprise AI will focus on how well organizations manage the conversations their AI models create.

Copilots are giving way to connected AI agents

During the first wave of enterprise AI, we embedded AI copilots into productivity tools, customer relationship management (CRM) systems and developer environments to improve how individual tasks were completed, and to help our employees work faster.

In the second wave, we’re deploying advanced AI agents in customer service, operations, sales and IT. These agents interact and communicate across voice, chat, email and messaging platforms while connecting with enterprise systems — and, increasingly, with each other.

At this scale, much of the complexity resides in the interaction layer. Every new channel, agent and use case multiplies the number of AI conversations happening across the organization.

Uncoordinated AI does not scale productively; it fragments instead

Consider the following scenario: One of your customers starts a conversation on web chat, follows up by sending an email and escalates the matter to a voice channel. Each interaction is handled by a different AI agent with a slightly different tone and behavior, drawing on different knowledge sources. The answers that emerge are technically correct but inconsistent.

Now add compliance requirements, regional variations and brand expectations, as well as third-party AI models and multiple copilots in your various business units. The result is predictably messy: disconnected communication channels, an inconsistent brand voice, fragmented knowledge and gaps in governance.

• ALSO READ → Service desk: What happens when agentic AI replaces the ticket queue?

The missing layer: A control plane for conversations

This challenge should feel familiar, because cloud computing went through a similar phase. Early adopters focused on infrastructure, including computing, storage and networking. But what made cloud truly viable was the introduction of control planes — centralized layers that govern, secure and orchestrate cloud resources.

Enterprise AI is now at the same point. It needs a control plane for conversations, and this what we have developed with NTT DATA’s Agentic AI Communications Gateway.

As outlined in a new NTT DATA guide, Bringing enterprise control and intelligence to every AI conversation, the gateway connects channels and enforces policies consistently while intelligently orchestrating AI-driven interactions.

Rather than treating each AI agent or channel as a separate system, it brings them together into a coordinated whole. It combines GenAI, enterprise knowledge and configurable controls to harmonize voice, chat, email and digital communications, and it applies governance and compliance guardrails. Every interaction ends up reflecting the right tone, brand voice and intent.

Keeping AI conversational while delivering business value

A persistent challenge in AI is the perceived trade-off between natural interaction and enterprise control. Organizations want conversations that feel human; they must be contextual, fluid and responsive. They also require compliance, data security and auditability, brand consistency and responsible AI behavior across markets.

These goals can be aligned only with the help of a communications control plane. Without stripping away the conversational quality that makes AI valuable in the first place, our gateway enables:

• Real-time, context-aware language capabilities
• Configurable brand personas and behavioral guardrails
• Knowledge-augmented responses grounded in enterprise data
• End-to-end encryption and governance frameworks

The result is something new: AI interactions that are both human-like and enterprise-grade.

Coordinating the human–AI workforce

Because agentic AI is designed to complement the work of human employees, the most effective operating models facilitate orchestration between humans and AI.

In these models, AI handles high-volume, routine interactions. Human agents step in to deal with complex, sensitive or high-value scenarios, with AI supporting them in real time by providing context, recommendations and next-best actions.

Our communications gateway makes this level of human–AI coordination possible. It manages when and how interactions are escalated, and ensures continuity across channels. First-contact resolution rates go up and handling times go down, resulting in loyal, satisfied customers.

Efficiency without compromise

When communication is centralized and orchestrated, it significantly reduces duplication between AI models, integrations and workflows. Interactions are routed intelligently, knowledge is reused, and your communication channels are managed as a single system, not as silos.

This leads to lower operational costs in the contact center, a more efficient use of AI resources, faster deployment of new services, and the ability to scale easily to match peak demand.

Importantly, these efficiency gains don’t come at the expense of customer experience, which also improves in quality.

Built for a multiagent, multicloud reality

Many organizations already manage hybrid environments that combine internal AI copilots, third-party agents and multiple cloud ecosystems. In Microsoft-centric enterprises, for example, tools such as Microsoft 365 Copilot, Microsoft Foundry, Teams and Dynamics 365 are only part of a broader landscape.

Our communications gateway sits above this layer of the environment. It is flexible and technology-neutral, orchestrating interactions between models, platforms and channels without locking your organization into a single path. It can also support agent-to-agent communication, not just human-to-AI interactions, and integrate with your enterprise systems  — including CRM — and knowledge bases.

This is what makes the gateway approach durable. Being multiagent, multicloud and multichannel, it will keep adapting to your evolving ecosystem.

Are you ready for the next wave of AI?

This is not a theoretical shift. At Mobile World Congress 2026 in Barcelona, we demonstrated the Agentic AI Communications Gateway in more than 100 live demos.

A consistent theme emerged from the conversations we had there: Organizations are shifting their focus from what AI can do to how it should behave — consistently, securely and aligned with business objectives.

Your organization won’t get ahead simply by deploying more AI agents. Coordinating them is where you’ll gain the most value. Your ROI now depends on governing how they interact, what they know and how they represent your business on every channel.

The communications control plane now becomes an architectural must-have. Make it part of your next wave of AI success.

• ALSO READ → Future of Work 2030: Making AI work in the workplace