Hacktoberfest 2026 · Touch Grass
Your next adventure starts outside.
Trade endless scrolling for real-world discoveries. Get a tiny mission, step outside, observe the world, and come back with something new.
“Your mission is outside. This screen is only the starting point.”
How FieldSkill works
A mission cycle, not a chat feed
- 01
Get curious
Pick a field mode. Gemma writes a tiny, safe mission.
- 02
Step outside
Read it once. Put your phone away.
- 03
Observe something real
5–15 minutes of looking, listening, counting.
- 04
Return with a discovery
Share what you saw. Get feedback and a better question.
The problem
Why FieldSkill exists
Most AI products measure success by keeping you engaged. FieldSkill measures success by helping you leave the screen, discover something real, and return with a better question.
The idea
Why real-world learning matters
Noticing, comparing and guessing causes are the roots of science. You can't learn how a shadow moves by reading about it — you have to watch one.
The model
Why open-weight AI (Gemma)
Gemma's weights are openly available, so the same model can be inspected, self-hosted or eventually run on your own device. This MVP calls Google-hosted Gemma remotely; the code is built so the provider can be swapped.
Your data
Privacy & responsible AI
No account. No location. No analytics. Your journal stays in your browser. Text is sent to the AI only after you agree, photos are never uploaded, and demo content is always labelled.
Honest limitations
- • A website cannot block other apps or detect that you went outside — completions are self-reported.
- • AI mission generation needs internet; saved journal entries work offline in your browser.
- • Photos are not analysed by AI in this version — feedback is based on your words.
- • Without a Gemma key, the app runs in clearly labelled Demo Mode.