After a week of using Nothing’s Essential Apps Builder for creating apps, a mixed impression emerged. The smartphone manufacturer’s vision for adaptive software is appealing, but its current implementation falls short. It appears challenging for this tool to evolve from a mere novelty into a dependable utility without significant improvements and sustained user patience.
Nothing previously introduced its concept for an “AI-native operating system,” intended to personalize and adapt its devices. This system, however, functions more as an AI layer over Android rather than a full operating system. It forms the foundation for Essential, described by CEO Carl Pei as the company’s overarching term for its AI products. Essential Apps, which are small, AI-designed widgets residing on the home screen, fall under this Essential umbrella. Due to their nature, ‘Essential Widgets’ might be a more accurate description.
These apps are created within the Apps Builder, located inside Playground, Nothing’s version of an app store. The process is straightforward: users describe their desired app in simple language, Builder generates it, and it can then be deployed to a phone. No complex setup or coding knowledge is required, though coding skills could potentially be beneficial. The Builder may ask for clarifications, and if the initial result is unsatisfactory, users can refine it iteratively instead of beginning anew.
Creating widgets that users would genuinely want to retain on their homescreen presents a different challenge. Even for someone who doesn’t typically use widgets, the significant difference between a functional widget and a truly useful one is immediately apparent.
Beginning with a simple request, a water-tracking widget was created to display a smiley face upon consuming eight glasses of water. While not aesthetically pleasing, it functioned correctly. An appointment widget, pulling data from a connected Google Calendar, was also easily created and worked on the first attempt. Subsequently, a small yellow mood widget was developed to show a different smile emoji with each phone unlock, later customized to blue. Updates were simple, involving adjustments in Builder and deployment to the phone. All projects were stored and organized in Playground folders, allowing for easy rollbacks.
Not every app created was a success.
However, the experience was not consistently smooth, especially when attempting to create more complex widgets. A shopping list widget, for instance, demonstrated the limitations of condensing app functionality into a small widget, displaying only a single item. Text truncation was also observed in several widgets. Location-based features proved problematic; a weather widget intended to use the current location instead displayed forecasts for four example London locations provided during setup, all within a single interface.
A poorly designed Pomodoro timer presented a more significant issue: it paused its countdown when the phone locked, rendering it ineffective for its intended purpose. Troubleshooting attempts were unsuccessful. Furthermore, a basic photo widget designed to display images from the camera roll failed to function entirely, and the Builder’s ‘fix with AI’ feature offered no resolution.
Upon reflection, two primary concerns hinder full adoption of Nothing’s vision for a dynamic ecosystem of AI-generated apps. The first stems from the product’s early beta stage. The Builder is currently restricted to Nothing’s Phone (3), supports only 2×2 and 4×2 widget dimensions, and offers full connectivity solely for location, contacts, and calendar data.
Nothing indicates future plans for apps to support a broader range of functions, such as internet data retrieval, media library and camera access, and Bluetooth device connectivity. New widget sizes, including compact 1×2 and larger 4×4 layouts, are anticipated around late March. Support for additional devices is also planned, with a public launch expected to introduce a wider array of user-generated apps. This is part of a new creator ecosystem that aims to allow users to “remix” existing apps. The exact timeline for these developments remains unspecified, with Nothing stating that a ‘public release will follow once system integrations are stable, permission handling is reliable and compatibility across devices is confirmed.’
The second, and potentially critical, challenge for such a project lies with the user. Experience with AI tools reveals a recurring pattern: regardless of a system’s capabilities, maximizing its potential is often the most difficult aspect. This challenge was immediately apparent when using Nothing’s Essential App Builder. While the tool appears highly capable and promising, users may struggle with defining their needs and articulating them effectively. An ecosystem founded on intuitive ‘vibe coding’ is an interesting concept, but intuition alone may not always suffice.
