Porchlight: a local AI outdoor coach that fits your body and your day
Description and use
Porchlight is a local AI coach that offers short, safe outdoor activities for people with mobility or sensory constraints, keeping all data on the user’s machine.
Problem
People with limited mobility, low vision, heat sensitivity, or only a few minutes cannot use typical outdoor apps that assume hiking or walking a mile.
Outcome
Porchlight enables short, safe outdoor activities for users with mobility or sensory constraints, keeping all data local and providing instant, personalized missions.
Workflow
- 1. Answer a one-time profile: max distance from home, stairs, heat sensitivity, available outdoor spaces.
- 2. Each session: input available minutes and current energy level.
- 3. Local Gemma model generates a JSON mission (title, place, minutes, distance, stairs, sun/shade, three steps).
- 4. Code validates the mission against user limits and a blocklist of unsafe words.
- 5. If valid, mission is displayed in terminal and as a large‑print HTML card; user performs the mission and returns to the app.
- 6. If invalid, the model is prompted again up to three times; otherwise a safe default mission is used.
Evidence
Porchlight is a small outdoor coach that runs on your own computer.
It runs an open-weight model (Gemma via Ollama) entirely on your own computer.
The model is Gemma (gemma3:4b), running locally through Ollama.
The app is plain Python using only the standard library, and it talks to Ollama's local HTTP API.
In my real runs with my own profile (150 m, no stairs, 15 minutes), Gemma followed the rules every time, so the guard never had to reject anything.
The model told me to stand barefoot on the doorstep. That can be unsafe for some people (hot or rough ground), and the guard let it through because it wasn't a rule yet.
- Record date: 2026-10-12T06:56:33.053Z