I Built an AI Gardening App That Tells Me to Put the Laptop Down

Description and use

Garden Trials lets gardeners systematically observe, test, and record plant changes using a local Gemma LLM and Streamlit, avoiding cloud services.

Problem

Gardeners often lack a structured way to observe, test, and record plant changes without relying on external tools or cloud services.

Outcome

Garden Trials enables gardeners to conduct controlled, repeatable observations and experiments on their own machine, keeping data private and avoiding cloud dependencies.

Workflow

• Write a note about a plant and the day you saw it.
• The Gemma model reads the note into plain facts and suggests one specific thing to check.
• You observe the plant and record the result before the app suggests a change.
• Choose one change from a short fixed menu or leave the plant unchanged for comparison.
• Count and optionally photograph the plant; the app shows the last photo for angle consistency.
• The app outputs a table, chart, verdict, evidence level (0‑3), and a plain next step.
• Save the trial data locally; the app guards against model errors and rejects unreliable outputs.

Evidence

Garden Trials is a small app that runs entirely on your own computer.

Source

The stack is small on purpose: Streamlit, one JSON file, and gemma3:1b running locally through Ollama.

Source

Setup is in the README: Python 3.11 or newer, Ollama, ollama pull gemma3:1b, then python -m streamlit run app.py.

Source