Conversational user interface: how AI makes software more personal

The way people interact with software has fundamentally changed over the past decades. From command lines to graphical interfaces filled with buttons, dropdowns, and forms. Ever more intuitive, ever more visual. But the next step is different in nature. Software is becoming something you talk to. A conversational user interface, or CUI, uses natural language as its main form of interaction. And since the breakthrough of modern AI, this is no longer an experiment, but a production-ready way of working.

In this article, we explain what a conversational user interface is, why it’s the future, and how we use them together with our clients in custom software.

Author: Allard Rothengatter

Published: 22 May, 2026

What is a conversational user interface?

A conversational user interface (CUI) is an interface that lets you communicate with a system using natural language. This can be via text, like a chatbot, or voice, like a voice assistant. Instead of clicking through menus, screens, and forms, you simply ask your question and get an immediate answer or see an action performed.

The evolution of interfaces follows a clear path. First, there was the command line interface (CLI), where users had to type precise commands. Then came the graphical user interface (GUI) with windows, icons, and menus. Now, we’re at the start of the third phase: the CUI, where software adapts to the user instead of the other way around.

The numbers show this is no small shift. According to various market analyses, the share of AI-supported customer interactions is expected to grow significantly in the coming years. Businesses are increasingly implementing conversational AI for customer service, sales support, and internal processes. Analysts also expect AI-driven solutions to help organizations save billions in operational contact center costs worldwide. At the same time, the conversational AI software market continues to expand rapidly due to the growing demand for scalable and personalized digital interactions.

The best interface is the one that needs no explanation. It’s not about whether a user understands the software, but whether the software understands the user.

– Allard Rothengatter

Why conversation works where menus fail

Traditional software assumes that users know where to go. You open an application, click through menus, learn what a function is called, find the right page, and fill in the right fields. That works fine for experienced users. But for anyone who only knows their problem—not the solution—it’s a struggle.

With a conversational interface, that logic is flipped. The user describes the problem and the system translates it into the right action. "I want a meeting room for Thursday afternoon for six people with a whiteboard" is much faster than clicking calendar, then location, then filtering by capacity, then filtering by facilities. The system takes over the thinking that used to be up to the user.

This brings three clear advantages over traditional interfaces:

Lower barrier to entry. No one needs to learn how the application works. If you can ask a question, you can use the software;
Context retention. A well-designed CUI remembers what was said earlier. That prevents repetition and makes each next interaction faster;
Personalization at scale. Answers adapt to who is speaking, what happened before, and what the system knows about the user. Previously, this was only possible with human contact.

Conversational interfaces in practice

The examples below have one thing in common: they don’t replace people, but they do replace menus. They are extra layers that make complex software more accessible for anyone who doesn’t have time to learn everything.

Microsoft

Copilot and GitHub Copilot make productivity and development tasks accessible through plain language. 88% of GitHub Copilot users say they are more productive thanks to this assistant.

ABN AMRO

Anna handles over 3.5 million customer conversations every year via text and voice. Customers manage their banking in plain language: from checking balances to making transfers and getting product support.

Duolingo

Duolingo uses a chatbot to guide users through language lessons with realistic dialogue exercises.

Anthropic & OpenAI

Claude and ChatGPT have become the starting point for many knowledge workers for research, writing, or problem-solving.

Conversation design: the discipline behind a great CUI

Building a chatbot isn’t technically hard. Building a chatbot that truly helps people is a different story. That’s the craft of conversation design, which combines UX, linguistics, psychology, and software architecture.

Good conversation design takes into account principles like:

Intent over keyword. A user who types "I can’t log in" doesn’t want to be sent to a page saying "keyword 'login' not found". A well-designed CUI understands the intent behind the question;
Remembering context. What a user said three messages ago should still matter three messages later. Don’t restart every conversation from scratch;
Clear fallback. If the system doesn’t know the answer, it should say so and refer the user to a human. Many chatbots lose trust by muddling through for too long;
Tone and personality. A good CUI has a voice that fits the brand. Formal for an insurance company, playful for an educational app;
Brevity. Three sentences beat three paragraphs. Users scan, they don’t read.

The impact of poor conversation design is huge. Research shows that more than half of users have been disappointed by a chatbot that couldn’t solve their question. That’s not an AI problem, but a design problem. With the right conversation design UX, that frustration disappears and only the benefits remain.

You can build a chatbot in an afternoon. Building a chatbot that truly helps people is a different story.

How we use CUI for our clients

At Dolphiq, we’ve been working with AI in our custom projects for years. For more and more clients, that means adding a conversational layer on top of their existing systems. Three examples from our practice:

Content

Answers from multiple sources

Conversational search on content pages

For one client, we built a chat interface that searches through various content pages and compiles a single concrete answer from multiple sources. No more lists of search results where the user has to decide what’s relevant.

The user asks a question, and the system determines which pages and sources are relevant, combines the information, and formulates the answer. With source references, so the user can click through for the full context.

The result: visitors get the right information faster, without having to scroll through multiple pages. AI does the compiling, we design the conversation and oversee the architecture around it.

Actions

An interface that takes action

From importing data to sending invitations

For another client, we built a conversational interface that not only answers questions but also performs actions and collects data. A chat as a control panel for the entire system.

Through chat, a user can send invitations for a test, compare test results from different people, or import data from various sources. Upload an Excel file, link a Google Sheet, or simply paste a piece of text into the chat: the system extracts the relevant data and processes it right away.

The difference from a traditional application: the user no longer needs to know where every function is. They say what they want, and the CUI translates that into the right actions. AI handles the interpretation, we design the workflow around it.

Portals

A natural layer on custom software

Conversational added to Qore.works

Many of our clients have extensive online portals with hundreds of functions, running on our Qore.works platform. Powerful software, but not every user finds their way through so much functionality right away.

By adding a conversational layer, users can simply ask for what they’re looking for. The CUI translates that into the right action within the existing system, using the API integrations already in place.

For the end user, it feels like a new, intuitive application. Under the hood, the existing software is still running, but now unlocked through language.

In all these projects, the same principle applies as with our other custom projects: AI does the heavy lifting, we design the experience. A CUI must fit business processes, the brand’s tone of voice, the legal context, and the users who work with it. Those considerations aren’t technical—they’re design questions.

Why this is the future

The numbers are convincing, but there’s a deeper argument too. For years, users adapted to software. They learned where the buttons were, what terms meant, and which screen order to follow. That’s not efficiency—that’s overhead.

The next generation of software adapts to the user. You tell the system what you want, and it figures out how to do it. For customer service, internal tools, search, onboarding, and support. For anything that used to require a manual.

That doesn’t mean traditional interfaces will disappear. For data-heavy work with lots of parameters, visual dashboards, forms, and maps will remain essential. But as an access layer, as the first interaction with software, the conversational user interface will become the standard. Just like the search bar became standard on websites over the past fifteen years.

For our clients, the strategic argument is simple: those who start thinking now about how users talk to their systems will soon have software that’s more intuitive than the competition’s. And those who wait until 2027 risk falling behind.

Conclusion: talking to software is becoming the norm

The conversational user interface is no longer just a feature. It’s a new interface paradigm. Just as the touchscreen transformed mobile software in the early 2010s, AI is now transforming how we interact with software at every level of organizations.

For our clients, this means a new layer on top of their existing custom software. A layer that is more accessible, more personal, and more intuitive than the menus and forms we’ve been building for twenty years. AI makes this layer possible. Good conversation design AI makes it usable. And experience with complex business processes makes it maintainable.

Curious if a conversational user interface would fit your application? Get in touch, we’d love to brainstorm with you.

Frequently asked questions about conversational user interfaces

A chatbot is a type of CUI, but not every CUI is a chatbot. A voice assistant is also a CUI, and modern AI assistants do much more than just answer questions. They take action, connect with systems, and execute complete workflows. So, a chatbot is a specific form within a broader field.
Conversation design is the craft of designing interactions between humans and systems using natural language. It combines UX, behavioral science, linguistics, and software architecture. A skilled conversation designer considers tone, fallbacks, context retention, and the path to a successful interaction. It is its own discipline, with unique principles.
With the right architecture, it is. Key considerations include data access, fallback mechanisms in case of uncertainty, logging, and auditability. We adhere to the principles of ISO 27001, the international standard for information security, and are currently undergoing the certification process. This means that security and privacy are integrated into every project from the very beginning.
Behind the scenes, a combination of Large Language Models (LLMs), natural language understanding, vector search for retrieving relevant context (RAG), and integrations with existing systems via APIs are at work. The choice of model, context architecture, and data integration determines whether it will be effective or not.
For open-ended questions, supportive tasks, and simple searches, a CUI works better. For data-intensive dashboards, configuration, and complex workflows, a visual interface is often more effective. Many modern applications combine both, using a CUI as an access layer on top of a richer visual foundation.

Software Development in the Age of AI

20 Mar 2026 by Allard Rothengatter

Much has been written in recent years about how AI is accelerating software development. And it’s true. Tools like Claude, GitHub, Copilot and other AI-driven environments reduce the time needed to write code, set up test scenarios, and handle repetitive tasks. But for us at Dolphiq, that’s not the most interesting part of the story.

We’d rather ask a different question: what will you do with the time AI gives back to you?

Vergelijking tussen standaard software en maatwerk software oplossingen

Custom Software Instead of Standard Software

6 Feb 2026 by Alan Linnenbank

You’re looking for new software for your company. The choice seems simple: a tried-and-tested standard solution used by thousands of businesses, or having custom software developed. The first option feels safe and straightforward, while the second might seem risky and expensive. But is that perception actually accurate?

In practice, at Dolphiq we often see organizations struggling with standard software. They start with the best intentions but soon run into limitations. In this article, we explain why custom software is not only a better choice for many companies, but often the smarter long-term investment as well.