Fieldnote: a tiny local AI prompt to help you notice more outside

Description and use

Fieldnote is a pocket‑sized web app that generates sensory prompts from short field notes, running a lightweight open‑weight model locally via WebGPU with a hosted fallback for unsupported devices.

Problem

Users want a quick, privacy‑respectful way to generate sensory prompts from brief field notes while walking, without sending data to external AI services.

Outcome

Fieldnote enables quick, privacy‑aware sensory prompts using a lightweight local model, with a fallback to a hosted model for broader device support.

Workflow

• Notice something outside
• Write a short field note
• Choose a route (local or hosted)
• If local and WebGPU compatible, run WebLLM + Qwen2.5-0.5B-Instruct-q4f16_1-MLC in browser
• If hosted, send note to Supabase Edge Function, which forwards to Backboard/Google Gemma via OpenRouter
• Receive one‑minute field prompt
• Put the phone away

Evidence

Fieldnote is a pocket sized outdoor reflection companion. Write one thing you noticed on a walk a bird call, a patch of moss, wind in the trees and it gives you one short sensory prompt.

Source

The local path runs an open-weight model in the browser. If a device cannot run WebGPU, Fieldnote still offers a clearly labeled sample prompt; it does not pretend that the sample came from AI.

Source

The hosted fallback is implemented, but I have not yet verified a complete live generation with its production provider credentials; I’m not presenting that route as a confirmed working demo.

Source