My Predictions for Conversation Design: 2026 and Beyond

As we look toward 2026 and beyond, I see several major shifts coming that will fundamentally change conversation design and how we work as conversation designers. These aren't just incremental improvements. These are the changes that will reshape our field.

Let me be clear about something: people saying LLMs can replace conversation designers? That's laughable.

I think people who say that don't know what conversation designers do. If all we did was write verbiage they may be right, but that's just the tip of the iceberg.

Conversation designers do very different work from company to company, which makes our role harder to define and easier to misunderstand. That's exactly why upskilling and understanding where this field is heading isn't optional anymore. It's essential.

CDI’s take: Conversation Design is a rapidly evolving space, and that means the role of Conversation Designer is moving too. What is becoming clearer than in previous years is that actively designing the interaction between human and AI is more critical than ever - precisely because LLM’s are powerful, yet unpredictable. 

I've watched IVR systems get converted to AI. The jobs didn't disappear. They transformed. And the people who saw it coming? They're the ones getting hired now.

The conversation design work isn't going away. It's evolving into more specialized, more technical, and frankly more interesting roles. Here's where I see the titles going:

Knowledge Engineer will be a title of the future. This is about structuring how knowledge is organized for AI, flows, taxonomies, intents, knowledge graphs. UX designers already create information architecture for apps and sites, this is information architecture for AI knowledge systems.

Model Designer is essentially designing the behavior of the model, defining how an AI model should behave in a product, tone, risk boundaries, consistency. This is designing the behavior of the model, akin to interaction design. Works closely with researchers.

AI Orchestration Designer is the big one. This is designing how multiple AI agents, tools, and systems coordinate to solve complex user tasks, the traffic controller for AI. UX designers think in systems and flows, this role is about mapping orchestration patterns, handoffs, escalation, tool use into usable experiences.

Multimodal Designer creates interactions that combine chat, voice, gesture, and visuals powered by AI. UX/UI designers are already skilled at multimodal thinking, this is the natural evolution.

I don't think conversational design is going away. I think it's going to evolve over time. 

CDI’s take: Although role terminology hasn’t caught up with the rapidly changing space, we fully agree that Conversation Designer is no longer a narrowly defined role. It touches on different capabilities - which is actually great, because it allows people to follow their expertise.

First: what is adaptive UI? Adaptive UI is an interface that automatically adjusts to each user’s needs, and it’s amazing because it makes everything easier, faster, and more personal.Here’s a clearer, more polished version:

“Here’s something that might surprise traditional UI/UX designers: conversation designers may actually build adaptive UI faster, because the workflow is much closer to what they already do.”

What's interesting about adaptive UI is that it's a bridge between now and when we get closer to “zero click Internet”. It is something we're going to see take off in the next few years. Most people haven't even heard of adaptive UI at this point, and basically what it is is the interface of the page builds itself based on your data and your history and your patterns and your inputs at the time and builds itself around what should potentially be the best possible experience.

Designing adaptive UI, if you've done conversational design, isn't a stretch, but if you've only done UI design or UX design, it will be a new skill set. We're already thinking about dynamic flows, context switching, personalization, and handling ambiguity. That's exactly what adaptive UI requires.

Now let's talk about what's really going to change the game: swarms and agentic systems.

I think designing orchestration will be the big one, as well as optimization expertise.

Swarms are complex. There's the queen bee and the agents and they all have jobs to do and relationships to each other like any other team. All of that will need optimizing.

CDI’s take: Multi agent or collaborative setups are indeed becoming a powerful trend. And once you understand how LLM’s and particularly RAG agents work, you will understand why it’s a good idea to define different sub-expert agents to solve customer needs. Even if, to the customer, it is invisible that they are being helped by a collection of domain experts. Swarms refers to a further step into an almost free-form collaborative space, by allowing simple, similar agents that follow basic rules to create a collective intelligence. Akin to ants or bees in nature. 

I don’t think the pre-swarm stage of optimization has fully matured yet, but once we’re optimizing actual swarms, things will become far more interesting, complex, and demanding.

Right now, we're still figuring out how to optimize single agent experiences. But when you're dealing with multiple agents working together, the optimization problem explodes in complexity. You're not just optimizing one conversation flow, you're optimizing an entire system of conversations that need to work in concert.

This is where conversation designers with technical depth will thrive. Understanding how to measure success across multiple agents, how to identify bottlenecks in orchestration, how to optimize for both efficiency and user experience, that's the expertise that's going to be in high demand.

CDI’s take: in the age of GenAI/LLM’s it’s easier than ever before to spin up an agentic solution that’s 90% there. But a huge part of the effort is now being spent getting that last 10% in order. It’s a radical shift from the way we used to design and build declarative solutions, but it is revealing that overall, there is the same amount of effort involved in producing a best in class solution. And, optimization is where it’s at!

People keep asking me the same question: "What happens to conversational design jobs when AI gets better?" I understand, but the better question is: "What happens to conversational design during the AI evolution?" Because that's where we are and it's going to take years.

Right now we're in the hybrid hell phase. Companies are retrofitting conversational AI onto click-based apps and both experiences are mediocre. Jobs are available for fixing broken hybrid interfaces, conversation design retrofits, AI experience designers, it's a vague title but there's real demand.

In 2 to 4 years, voice-first becomes default, screens become fallback. Navigation dies, prediction takes over. Jobs shift to conversation designers, prompt engineers for UI, multimodal experience designers, failure mode specialists.

In 4 to 7 years we'll see the new specializations. Ephemeral UI design, interfaces that appear only when needed. Context-aware prediction systems. Personal AI customization. New roles like AI personalization designers, serendipity engineers, yes really, verification UX specialists, and orchestration designers managing complex agent swarms.

CDI’s take: We applaud Anne’s bold predictions. Conversational AI is a rapidly evolving field and we are both curious about and fascinated by the developments in the market. By keeping our finger on the pulse of change, we strive to keep our students and clients ahead of the field!

The designers surviving this aren't the ones with the best chatbot portfolios. They're the ones who understand how conversation fails differently than clicks, can design for ambiguity not binary actions, know when a form beats a chatbot, get comfortable with AI as a design material.

You may not need to learn to code but you need to learn the branches of AI, how they work in design. Be able to be multimodal designers and understand systems and how they work.

In regard to design, I think the word AI design is not going to stick. I think AI is going to taper off being in the headline and just sort of be. I think a lot of people are predicting that, that's nothing groundbreaking.

What is groundbreaking is understanding that conversational design is evolving into something more specialized, more technical, and more essential than ever before. The LLM can generate text. It cannot design the system that text lives in. It cannot design the orchestration of multiple agents. It cannot optimize for user experience across complex workflows. It cannot make the strategic decisions about when to use conversation and when to fall back to UI.

That's our work. And there's more of it coming, not less.

I'll say it plainly, I've watched IVR systems get converted to AI. The jobs didn't disappear. They transformed. And the people who saw it coming? They're the ones getting hired now.

The question isn't whether conversational design has a future. The question is whether you're preparing for what that future actually looks like.

CDI’s take: We couldn’t agree more with this powerful conclusion. Conversation design is here to stay as Conversational AI really starts to deliver on its promise: an interaction between human and machine using our preferred method of communication: language.