Pocket Field Trip: local AI with somewhere to stop

Description and use

Pocket Field Trip guides users on short walks with AI‑selected observation prompts, runs inference locally in the browser, and stores notes offline, reducing screen time and data usage.

Problem

Users want a simple outdoor observation task without constant screen interaction or reliance on remote AI services.

Outcome

The app runs inference locally in the browser, matching user input to 20 observation cards using 384‑dimensional embeddings, and provides a regression‑tested note‑saving flow.

Workflow

• 1. Open the browser app or Android build.
• 2. Choose a prompt from the observation card list.
• 3. Read the observation card and step away from the screen.
• 4. Return to the app to save a short note about what was observed.

Evidence

The project uses an open-weight text embedding model through Transformers.js and ONNX Runtime Web. Inference runs locally in the browser through WebAssembly.

Source

The published demo has been tested with real browser inference, producing 384-dimensional embeddings across a collection of 20 cards.

Source

The browser version is ready to try. Android testing is still based on an emulator, so I am not claiming broad physical-device validation.

Source