The revolution of AI first interfaces

Everyone is talking about AI. And I'm also a big believer that this leap forward in technology is here to stay and will fundamentally change many aspects of how we live and work. In fact, it is already happening. One interesting thought that I had the other day, and found worth writing down and expanding on, was the change of software interfaces. Here is my take I want to share with you.

At the intersection of software development and AI, not only the way how we build products has completely changed, but also the way how we interact with applications.

The last huge, pivotal moment in modern software history was definitely the change from command-line interfaces to graphical user interfaces (GUIs). Back then, it was required to know and understand computer language to let the machine do things for you. With GUIs, we have catapulted ourselves into a time of making software globally accessible and giving everyone the power to use programs.

Now is the time for the next revolution of app interfaces. AI-driven interfaces that transcend efficiency, evolving into empathetic collaborators. This shift isn’t merely technical – it’s a redefinition of human-machine relationships. So let's unpack what AI first interface means.

Early AI tools focused on completing tasks – think about Siri and Alexa, which took natural language processing (NLP) inputs and were barely just good enough to do things like setting timers or getting asked for the weather report.

With breakthroughs in large language models (LLMs), things changed rapidly. Suddenly things like context-aware computing, reasoning, semantic search and natural language processing (NLP) got so much better that they led to better user experiences and opened up new possibilities for how we interact with software. Things that were previously unimaginable have now become reality. And I think we are only just beginning to exploit the capabilities of AI. Whether and when we reach artificial general intelligence (AGI) is a question for another time.

One of the most famous examples that reached widespread adoption in society is obviously ChatGPT. In general, chat interfaces have advanced so much that even long-established state-of-the-art things like googling something are being disrupted. It has almost become a habit for many to ask a question to ChatGPT, Perplexity, Claude, DeepSeek – you name it – AI models to get the answer more seamless. The way clicking through some links and websites that Google presents you with sprinkled in ads and lots of distraction and SEO optimized content just feels outdated. It feels like you have the accumulated knowledge of the entire web at your fingertips.

Use cases such as customer support via chatbots or phone calls finally work without the need of a human in the loop. Crafting content in written, image, or soon also video form has gotten incredibly good. So good that it becomes almost impossible to tell what is AI-generated and what is not. Not going to cover ethical topics in here, but certainly an important topic.

In general, I think this shift reflects a broader trend: technology and software becomes much more your partner rather than a tool with hard-coded functionality. From a technical perspective AI creates an infinite amount of different user flows that are possible because you don't have a pre-defined path that every user has to take. The interactions with AI, especially in conventional applications, are endless.

Software that understands your intent, has much more context, and can even feed in your previous interactions to learn from it and adopt specific behaviors that are tailored to an individual user is just around the corner.

But not everything in this AI interface revolution is just about chat interfaces. It doesn't make sense all the time to replace every button click with a voice or text command. In some cases, it is still faster and more seamless to click a button instead of using AI. So I don't believe that the future of software interfaces is just a simple chat interface. Different tools for different use cases will always need some form of customized interface to display the data and information that should be presented. But I believe they will become more dynamic and adapt depending on how the software is used. I think about a kind of proactive intelligence that can bring the best possible user experience.

If we look at the latest technical trend, agenetic workflows (AI agents), we see exactly this transformation. For no subject-matter experts, it is hard to grasp how exactly these AI agents work. In a very high level abstraction, you can think about LLMs with the capabilities of understanding complex inputs, engaging in reasoning and planning, using tools reliably, and recovering from errors. Instead of setting up specific workflows that app developers bake into their AI applications by using multiple prompts chained together to deliver a great output for the user, an AI agent plans and operates independently. It can take many turns and iterations to find the path that delivers the right outcome, or pause the execution to get necessary inputs from the user to overcome blockers and assess its progress.

I hope this short excursion was exciting and has inspired you to think about how app interfaces change. Personally, I find this topic incredibly exciting because it gives us product developers so many opportunities to go back and think first principles and to question and rethink every traditional interaction between humans and software.