All manner of companies are rushing to integrate AI into their products, but simply adding AI to existing workflows might not be the way to go.
Pete Koomen, a Partner at the influential startup accelerator Y Combinator, has offered a compelling perspective on the integrating with AI, cautioning developers against falling into the “AI horseless carriage” trap – a phenomenon where new technology is superficially applied to old paradigms, limiting its transformative potential.
Koomen’s central argument is that many current attempts to integrate AI into software mirror the early days of the automobile — in a suboptimal way that attempts to insert AI within existing frameworks, instead of rethinking applications in the age of AI. He explained this through the concept of “horseless carriages”. “This is a reference to early automobile designs that looked a lot like carriages with the horse replaced with an engine,” Koomen explained.
He elaborated on the flaws of this early approach to car design: “There were all sorts of problems with that design. For example, there’s less suspension on carriages, which didn’t work when running at high speeds with a vibrating motor. The center of gravity was higher, which made turns harder at high speeds. Basically, inventing the motor was only a small part of what was needed to produce a vehicle that could take advantage of the motor’s power. It only became useful once you redesigned the entire vehicle.”
This pattern, Koomen argues, is not unique to automobiles but repeats throughout technological history. “This phenomenon, we see over and over and over again in technology,” he stated. “Some ones that I have lived through: when the internet came about, a lot of the first search engines were literally just digitized yellow pages. It was just a directory of listings. And now, of course, we think that’s silly. Craigslist is kind of what it started with.”
Fellow Y Combinator partner David Lieb gave another example. “When mobile came out, the first mobile apps, or a lot of the first mobile apps, were basically just websites wrapped in a native app wrapper. They didn’t take advantage of any of the new technologies available on the mobile phone, like GPS or multi-touch. It takes a few years typically to get to the useful bit of the new technology,” he said.
Applying this to the current AI landscape, Koomen suggested that we are still in the early, imitative phase. He pointed to common practices in major applications as evidence: “I think the deepest problem here is that when the Gmail team set out to build [AI features], they kind of asked, ‘How can we slot AI into the Gmail application? How do we replace the horse and put an engine in?’ That’s exactly right.”
The issue with this approach, according to Koomen, is that it fails to address the core potential of AI. “The problem with that is that Gmail is an application designed for a human to do work in,” he explained. “The real promise of AI, I think for many of us, is using AI to automate repetitive busy work. A lot of the time I spend on email is repetitive busy work. It’s work that doesn’t really need my full brainpower, but because we haven’t had the technology, it requires it. And so I give a little example in my essay of just, you know, using these simple techniques, what you could do with an email.”
Beyond the Horseless Carriage: Reimagining AI in Software
Koomen’s “horseless carriage” critique serves as a vital warning for the tech industry. It suggests that simply embedding AI features into existing software architectures—like adding a chatbot to a legacy CRM or an AI summarizer to a traditional word processor—may only scratch the surface of AI’s capabilities. The real transformation, as history shows, comes from fundamentally rethinking products and workflows around the unique strengths of the new technology. This means designing “AI-native” applications where the AI isn’t just an add-on but a core component that enables entirely new ways of solving problems and delivering value.
The implications are profound. For developers and product managers, it necessitates a shift from asking “How can AI improve our current product?” to “If we were to build this product from scratch with AI at its core, what would it look like?” This could lead to radically different user interfaces, automated processes that were previously unimaginable, and solutions that proactively address user needs rather than reactively responding to commands. For businesses, embracing this AI-native mindset could unlock significant competitive advantages, driving efficiency and creating novel user experiences that legacy systems cannot match.
Navigating Towards Truly AI-Driven Applications
The journey from “horseless carriage” AI to truly transformative, AI-native applications is ongoing. We see glimpses of this future in emerging AI agents that can perform complex multi-step tasks, generative UIs that adapt to user needs in real time, and platforms that use AI to automate significant portions of creative or analytical workflows. However, overcoming the inertia of established practices and the comfort of familiar software paradigms requires deliberate effort and a willingness to experiment.
The challenge lies in identifying and shedding the “carriage” aspects of our current software – the design choices and limitations inherited from a pre-AI era. As Pete Koomen’s analogy powerfully illustrates, the engine of AI has arrived. Now, the task is to design the entirely new vehicles that can truly harness its power, moving beyond mere imitation to genuine innovation. This will require not just technical skill, but a visionary approach to product development that prioritizes what AI makes possible over what traditional software has always done.