Trail Quest: Private, On-Device AI for Nature Walks
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass What I Built Trail Quest is a mobile-first web app that turns a little time outdoors into a nature-spotting quest. Choose a setting—such as a park, garden, neighborhood, trail, or balcony—and a time limit. Trail Quest gives you three observation challenges, then encourages you to put your phone away and…
Trail Quest is a mobile-first web application that transforms outdoor time into a nature-spotting adventure. Users can select a location—such as a park, garden, neighborhood, trail, or balcony—and set a time limit. The app then presents three observation challenges before prompting users to put their phones away and explore. Participants can choose to complete or skip challenges, keep private reflections, and save completed walks in a local nature journal. The app also features a high-contrast sunlight mode for outdoor use.
The development of Trail Quest relies on WebLLM to run the open-weight SmolLM2-135M-Instruct-q4f16_1-MLC model directly in the browser using WebGPU. This enables quest generation to occur on the user's device, ensuring that the location and prompts are not sent to an AI API. In cases where the model cannot run on a device, Trail Quest offers handcrafted quests that can be used offline.
As for the AI mode, it requires an initial model download; however, the browser can then reuse its cached model for local generation, provided that the browser supports WebGPU and has sufficient storage.
The user interface of Trail Quest is structured around a simple, accessible flow: choosing a setting and duration, receiving three quests, and then transitioning to a "phone away" screen. The app also incorporates safety guidance, local journal storage, Markdown export, and synthesized chimes using the Web Audio API.
The open-source nature of Trail Quest is crucial because it ensures that personal outdoor plans remain on-device and do not get sent to a hosted AI service. This allows for model inspection and swapping, and the handcrafted fallback ensures the app remains functional when AI inference is not available. This is particularly important for an app that aims to encourage individuals to pay attention to the world around them.
By keeping personal inputs on-device, reducing dependence on paid APIs, and enabling community adaptation for various models and devices, Trail Quest exemplifies the value of open innovation. Render hosts the Trail Quest web app as part of the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
