{"slug": "field-break-an-open-weight-ai-that-tells-you-to-close-the-screen", "title": "Field Break: an open-weight AI that tells you to close the screen", "summary": "A developer built Field Break, an open-weight AI tool that returns a single outdoor micro-adventure and then instructs the user to put the screen away, with the goal of minimizing interaction time rather than extending conversation. The project runs on the Apertus 1.5 model (default swiss-ai/apertus-v1.5-8b) through an OpenAI-compatible endpoint, with provider and model configurable via environment variables and a clearly labelled deterministic demo-policy fallback when no API key is set. A live Apertus 1.5 70B test produced a \"Neighbourhood Pocket Adventure\" ending with \"Turn off phone and let adventure unfold offline,\" and the exact prompt and result are preserved in the repository.", "body_md": "*This is a submission for the [Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass](https://dev.to/challenges/hacktoberfest-week1-2026-10-05).*\n\nI built **Field Break**, a deliberately small AI tool whose success condition is that you stop using it.\n\nYou tell it four things:\n\nIt returns **one** micro-adventure that can start almost immediately, plus a final instruction telling you when to lock the screen.\n\nThat last part is the product idea. Most AI apps optimize for more conversation. Field Break optimizes for the shortest useful interaction possible.\n\nThe 36-second demo is here:\n\n[https://raw.githubusercontent.com/ondmindmanagement-hub/field-break/main/demo/field-break-demo.mp4](https://raw.githubusercontent.com/ondmindmanagement-hub/field-break/main/demo/field-break-demo.mp4)\n\nA live Apertus 1.5 70B test was also captured during the challenge. Given 20 minutes, medium energy, neighbourhood streets and a small park, the model produced a “Neighbourhood Pocket Adventure” and ended with:\n\nTurn off phone and let adventure unfold offline\n\nThe exact prompt and result are preserved in the repository so the model evidence is inspectable rather than described from memory.\n\nHacktoberfest Open-Source AI Challenge — Week 1: Touch Grass\n\nField Break turns a small amount of free time into one simple outdoor micro-adventure, then explicitly tells the user to put the screen away.\n\nThe app is designed for the open-weight Apertus 1.5 model through an OpenAI-compatible endpoint. The provider/model are environment-configurable, so the project can switch to another open-weight deployment without changing the UI.\n\nRun with real model inference by setting:\n\nOPEN_MODEL_API_KEY\nOPEN_MODEL_API_URL=[https://api.publicai.co/v1/chat/completions](https://api.publicai.co/v1/chat/completions)\nOPEN_MODEL_NAME=swiss-ai/apertus-v1.5-8b\n\nWithout a key, the prototype runs a clearly labelled deterministic demo-policy; it never pretends that fallback output is model inference.\n\nThe planning layer is not locked to one proprietary model or API. An open-weight model can be self-hosted, swapped, audited, or moved closer to the user's data. For a tiny tool whose purpose is to get you away from the screen, that simplicity matters.\n\nField…\n\nRepository: [https://github.com/ondmindmanagement-hub/field-break](https://github.com/ondmindmanagement-hub/field-break)\n\nThe project is intentionally dependency-light: a Python standard-library server, a small browser UI, three unit tests, and an OpenAI-compatible adapter for an open-weight model endpoint.\n\nThe open model at the center is **Apertus 1.5**, with the default configuration pointing to `swiss-ai/apertus-v1.5-8b`.\n\nThe model receives only the context the user provides. The system prompt explicitly tells it not to invent live weather, trail closures, local conditions, or medical claims. It must prefer something local, simple, low-cost and reversible.\n\nThe output is structured JSON:\n\n`screen_exit` instruction.\nThat structure matters because I did not want another chat interface that can drift into an endless conversation. Field Break asks for one decision and then gets out of the way.\n\nThe provider and model are environment-configurable:\n\n```\nOPEN_MODEL_API_KEY=...\nOPEN_MODEL_API_URL=https://api.publicai.co/v1/chat/completions\nOPEN_MODEL_NAME=swiss-ai/apertus-v1.5-8b\n```\n\nIf no model key is configured, the app switches to an explicitly labelled `demo-policy`. It never claims that deterministic fallback output came from the model.\n\nI also added tests for JSON extraction, output normalization and the demo planner. All three are passing.\n\nFor this project, open innovation is not a decorative technology choice.\n\nA closed planning API would make the smallest part of the product — deciding what to do outside — dependent on one vendor. With an open-weight model, the inference layer can be moved, swapped or self-hosted without changing the experience.\n\nThat becomes especially interesting for a tool whose philosophy is **less cloud, less screen, less dependency**.\n\nThe project can use a hosted Apertus endpoint today, but the interface is deliberately compatible with a future local or privately hosted deployment. The user interface does not need to know which provider is behind it.\n\nOpen weights also make the model choice inspectable. The project is not pretending that “AI” is a magic black box; the model family, prompt, evidence and fallback behavior are all visible in the repo.\n\nThe challenge theme pushed me to reverse the normal metric.\n\nInstead of “How many messages can the user send?”, I asked:\n\n**How quickly can the software become unnecessary?**\n\nThat changed the design:\n\nThe best output is not the most impressive paragraph. It is the one that makes someone say “okay” and leave the desk.\n\nI am entering the **overall Hacktoberfest Open-Source AI Challenge Week 1** category.\n\nI did not add a partner category just to increase eligibility; the project only claims technologies it actually uses.\n\nField Break is small on purpose. The screen should be the shortest part of the experience.", "url": "https://wpnews.pro/news/field-break-an-open-weight-ai-that-tells-you-to-close-the-screen", "canonical_source": "https://dev.to/unfire/field-break-an-open-weight-ai-that-tells-you-to-close-the-screen-35j9", "published_at": "2026-10-05 22:55:45+00:00", "updated_at": "2026-10-05 23:18:06.289098+00:00", "lang": "en", "topics": ["ai-tools", "large-language-models", "generative-ai", "ai-products"], "entities": ["Field Break", "Apertus 1.5", "swiss-ai/apertus-v1.5-8b", "Hacktoberfest Open-Source AI Challenge", "GitHub", "PublicAI"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/field-break-an-open-weight-ai-that-tells-you-to-close-the-screen", "markdown": "https://wpnews.pro/news/field-break-an-open-weight-ai-that-tells-you-to-close-the-screen.md", "text": "https://wpnews.pro/news/field-break-an-open-weight-ai-that-tells-you-to-close-the-screen.txt", "jsonld": "https://wpnews.pro/news/field-break-an-open-weight-ai-that-tells-you-to-close-the-screen.jsonld"}}