Could AI code assist unlock a new era of web and mobile user experience?
Speculative ideas about AI agents, app design, and the end of “static” mobile UX
AI-assisted code development tools exist on a spectrum between code visibility and ease of use. Tools like Cursor and Windsurf occupy the code-forward side, appealing to those already familiar with coding environments.
On the opposite end, platforms like lovable.dev make source code accessible but secondary. They target users who want to design apps without engaging with code at all.
The remarkable success of lovable.dev revealed my own bias about app creation. I had assumed the creative drive to build apps belonged primarily to those who could code. I believed technical skills preceded inspiration, but lovable.dev demonstrates these elements are actually independent. Lovable.dev’s impact shows that a much wider audience beyond software developers wants to create apps. It’s effectively unlocking creativity for everyone.
How does a no-code dev environment evolve as LLMs improve?
The LLMs powering lovable.dev and similar ‘no-code’ app development tools will become increasingly powerful in both effectiveness and efficiency. They’ll generate better code while consuming fewer resources, eventually enabling LLMs to run on edge devices, particularly smartphones.
What happens when LLMs advance to where everyone can run a supercharged version of lovable.dev locally on their phones? Imagine a world where your smartphone has enough AI coding capability to create any app exactly as you want it, simply by describing your desired user experience.
If no-code development environments become highly capable, user-friendly, and run locally on devices, what new forms of creativity might this unleash?
ChatBot, meet ‘DevBot’
The next evolution beyond platforms like lovable.dev would likely be website and mobile app creation from within the app itself, using an interface similar to today’s chatbots.
On retail and commercial websites, you often see a chatbot in the corner answering customer questions. Now imagine instead a “DevBot” in that same corner, waiting to custom-build experiences based on your instructions. You’d simply click on the DevBot and describe what you want. Or, more likely, each user steadily customizes their user interface and functional particulars from some starting point template, supported by guardrails and limits.
If such a DevBot had sufficient power with deep OS integration through official APIs or on-device development sandboxes, and could access the content that fills today’s apps (social media, news, sports, streaming content), how might this transform our current website and mobile app landscape?
Why mobile operating systems might enable something like this
Let’s explore how a ‘DevBot’ concept could transform mobile user interfaces. Why would operating systems relinquish such control to a DevBot?
This question reminds me of the story in Walter Isaacson’s book about Steve Jobs’ initial resistance to opening the App Store to third-party developers. Jobs hesitated, doubting Apple’s ability to provide the necessary software tools, drivers, and libraries for app development.
Apple’s eventual release of an SDK and development tools for third-party apps unleashed a creative and business explosion that shaped today’s mobile experience dominated by third-party apps. This move also created an incredibly profitable revenue stream for Apple.
As no-code development environments like lovable.dev gain momentum and tools become increasingly user-friendly, it’s possible that companies owning phone hardware and mobile operating systems would find a way to natively incorporate no-code app development. If these companies establish effective monetization models, ‘DevBot’-style personalized control could fundamentally transform how we interact with our phones.
How content providers fit into this concept
We go online and use our phones to interact with and consume content largely owned by third parties. The amount of available online content is effectively infinite. From social media accounts to 24-hour streaming sports and news, games, and vast libraries of movies, books, music, and on-demand TV, we have endless ways to spend our time.
However, we navigate an online world of content consolidation. This consolidation serves a purpose: platforms like Facebook, Netflix, Amazon, and Google help us find signal in the noise. They curate the content we want and, in exchange, sell our attention to advertisers. This symbiotic relationship has benefited both companies and consumers.
If this era of “static” apps ever ends, I believe consumers gaining more control over how they access third-party content will drive that transformation.
This doesn’t mean content would become devalued or free. Content creators still need and deserve compensation. Rather, content providers might someday be compelled to cede enough control over their content to support deeply customized user experiences.
It’s a long shot, but if we gain more control over our feeds, the revenue models emerging in such an environment could be less exploitative and manipulative to both content creators and end users than the revenue models that drive technology today.
Onto some experiments
I plan to build a minimally functional proof of concept for a DevBot. I’ll begin with a DevBot-customizable web app template. My goal is to create something that can deploy website features that go beyond basic CSS edits. Let’s see if I make enough progress by next week to continue this series with a second part.