The app-less smartphone. A dynamic platform that harnesses natural language processing to generate adaptive interfaces in real time.
(0.1) Background
Brain.ai was an AI and interface company founded in 2015 with a simple but powerful vision: to organize the world’s software and make it human centered and natural to use. The company developed new technologies, interaction metaphors, and developer platforms that allow computers to feel less like tools and more like an extension of the human mind.
In 2016, Brain pioneered one-shot learning NLP, a breakthrough that influenced many of the language models that followed. Building on this foundation, the team launched Natural AI in 2020, the world’s first consumer-facing generative interface. Instead of static apps, users could simply express intent in natural language, and Brain’s engine would generate the interface in real time.
(0.2) The Process
I first encountered Brain.ai in 2016 through a mutual connection and followed their journey closely over the years, witnessing both their technical breakthroughs and their product struggles. In 2020, as the founders prepared to launch their app, I joined as a consultant to help shape the interaction model and user experience of what was, at the time, a first-of-its-kind platform. This was pre–chatGPT era, and the challenge was to design an interface that could respond intelligently and naturally to human language.
My role focused on architecting interaction systems that could operate fluidly through both voice and text. I collaborated closely with engineers and product architects to translate conversational input into structured interface states that were adaptive, fast, and intuitive. My years as a research aide and usability lab facilitator gave me a unique advantage here. I had learned to map subtle human speech patterns, pauses, and intent into design logic.
A key contribution was the development of voice-first interfaces. I designed adaptive voice flows that could handle incomplete, ambiguous, or interrupted input without breaking context. This required close collaboration with the NLP team to shape response timing, conversational feedback, and error recovery patterns. Instead of relying on fixed command trees, we built dynamic conversation models that responded in real time to user behavior.
For the interface layer, I worked in Figma to build high-fidelity prototypes, layered with interaction design patterns and motion states to simulate real-time response. These prototypes were stress-tested with multiple user scenarios to validate how well the language model could render and adapt the UI on the fly. We also integrated usability testing protocols to observe how users naturally spoke to the system, adjusting the conversational architecture accordingly.
Every decision was guided by a single principle: make technology feel invisible. The aim was to create a platform where language itself became the interface, closing the distance between thought and action.
Team Size
4
(0.3) The Results
Every decision was guided by a single principle: make technology feel invisible.
The outcome was an early prototype of a truly app-less smartphone experience. Instead of navigating menus or downloading multiple apps, users could simply speak or type their intent, and the system generated the required interface on the spot.
The platform proved that AI can collapse complex workflows into a single interaction, shifting the burden of adaptation from the user to the software. It showed a future where software listens, understands, and builds around human intent rather than forcing users to conform to predefined structures.
Brain.ai’s Natural AI became one of the earliest public demonstrations of real-time generative UI, long before ChatGPT reached mainstream adoption. It laid the foundation for a new class of interaction design in which human language functions as both input and architecture.