Fresh Air Match: local AI that gets you outside without tracking you

Description and use

Fresh Air Match uses a local MiniLM model to quickly and privately suggest three short outdoor activities, respecting user constraints.

Problem

Users want quick, privacy‑preserving outdoor activity suggestions based on mood and constraints.

Outcome

The app delivers three curated outdoor activities within user constraints, preserving privacy and ensuring deterministic safety filtering.

Workflow

• Collect user constraints (time, weather, company, energy, access, free‑text mood).
• Apply deterministic safety/access filters to a curated catalogue of 16 activities.
• Run local MiniLM embeddings on the free‑text note.
• Rank remaining activities by cosine similarity.
• Return top three activities with time bounds, plan, bring list, and safety reminder.

Evidence

Fresh Air Match turns a short description of your mood into three small outdoor activities you can actually do.

Source

It filters a curated catalogue of 16 activities using the hard constraints first. Only the eligible activities reach the semantic-ranking stage.

Source

A local sentence-transformers/all-MiniLM-L6-v2 ONNX model then compares the note with those activities and returns the three closest matches.

Source

The model runs locally through ONNX Runtime. Preference text is not sent to a hosted model, and the app asks for no precise location.

Source

Semantic similarity never overrides a hard constraint. The server first removes activities that do not fit the requested time, company, weather, setting, energy, or access requirements.

Source

I ran ten unit and integration tests covering: input validation and length limits; catalogue integrity; time, weather, company, setting, energy, and accessibility filters; semantic ranking; privacy and safety statements; and real ONNX inference when the model artifacts are present.

Source