WildQuest: Offline Nature Quests with On-Device Qwen AI

Description and use

WildQuest is an offline-first mobile game that generates nature exploration missions and processes photos using an on-device Qwen model, keeping all data local and private.

Problem

Users want structured nature exploration without relying on internet or cloud APIs, preserving privacy and enabling offline use.

Outcome

WildQuest enables structured, privacy-preserving nature walks entirely offline, with on-device AI mission generation and photo analysis, eliminating the need for cloud APIs and ensuring data stays on the device.

Workflow

• Select walk duration (15, 30, or 60 minutes) and theme (plants, birds, textures, general nature).
• Generate 3–5 missions using the on-device Qwen2.5-1.5B-Instruct Q4_K_M model via llamadart.
• Follow missions outdoors, taking photos and writing brief descriptions.
• Submit photo and description to the same on-device model to receive structured JSON (label, confidence, key features).
• Save notes and mission completions locally in SQLite; drafts survive app restarts.
• Review field journal summarizing completions, observations, and next walk questions.

Evidence

WildQuest is a mobile-only, offline-first nature exploration game. You pick a 15-, 30-, or 60-minute walk and a theme (plants, birds, textures, general nature), and the app generates 3–5 short missions from an on-device open-weight LLM — with a bundled curated mission library as fallback, so it works with zero connectivity.

Source

The point is the screen stays secondary: you read one mission, put the phone away, and go outside. When you come back you photograph what you found, describe it in a sentence or two, and the same on-device model structures it into an ID suggestion (label, confidence, key features).

Source

Everything — quests, photos, notes, AI text — lives in on-device SQLite, including half-typed drafts that survive app restarts.

Source

The open-source AI is the core, in three places: Quest generation — Qwen2.5-1.5B-Instruct Q4_K_M (official Qwen GGUF on Hugging Face, ungated, Apache-2.0), run on-device through llamadart 0.11 (llama.cpp), CPU, 2048 context.

Source

No server, no account, no API keys. A Python/FastAPI + Ollama backend exists in the repo purely as a dev reference with the same prompt shapes and rules.

Source