Flâner: An Open-Source AI Companion for Going Outside
Flâner is an open-source AI companion built for Hacktoberfest that generates short outdoor plans to help users step away from their screens without using points or streaks.
Stock photo for illustration only, not from the actual event
- Flâner is an AI companion designed to help users take small steps outdoors.
- It requires only about a minute of conversation before telling you to put your phone away.
- It features no XP, streaks, or leaderboards to prevent digital addiction.
- Built with vanilla HTML/CSS/JS frontend, Node/Express backend, and Supabase.
With comfortable homes and endlessly engaging smartphones, people frequently stay indoors and gradually stop noticing their surrounding streets, lights, and neighborhood shops. The hardest part of going out is almost always taking that initial step, yet the generic advice to "go outside more" remains too vague to act upon effectively.
Created as a submission for the Hacktoberfest Open-Source AI Challenge Week 1 under the theme "Touch Grass," Flâner was developed to address this exact friction. It functions as a gentle AI companion that translates a vague desire to leave the house into a concise, actionable checklist for a short outing.
Unlike conventional apps engineered to maximize screen time, Flâner is intentionally built to demand less attention. Users converse with the AI for roughly a minute, receive a small itinerary, and are immediately instructed to put their phones away. The sense of achievement comes entirely from the real world rather than in-app points or streaks.
Designing against the traditional engagement loop represents a counter-cultural approach in modern software development. While most applications fight for continuous user retention, Flâner adds value by actively encouraging users to disconnect from digital displays and reconnect with their physical surroundings, demonstrating a healthy use-case for AI in personal well-being.
Beyond planning, the app includes a reflection feature. Upon returning home, Flâner prompts users to recall what they observed and noticed along the way, training their attention for future outings. Outings that are started but left unfinished are never saved to the database, keeping the dashboard focused solely on completed achievements.
The creator tested the app after postponing a grocery run for three days, noting that having the shopping list pre-populated by the AI eliminated hesitation at the store. Simple preliminary steps—such as freshening up and getting dressed—helped build the momentum needed to complete the ordinary errand successfully.
Technically, Flâner utilizes a vanilla HTML/CSS/JS frontend without frameworks or build steps, paired with a Node and Express backend. It relies on Supabase for authentication and database management, deployed as a single web service on Render, utilizing open-weight gpt-oss-20b on Groq and Gemini for its core AI processing.
Source: Dev.to
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment