I gave an agent a lakehouse and control of my garden
Web page · https://jatorre.github.io/garden
First seen by Oriel on .
AI summary
The author turned a two-square-metre vegetable garden into an agent-monitored system combining sensors, camera images, forecasts, and scheduled analysis in a sovereign LakehouseBox deployment.
- Eleven Ecowitt devices collect soil, weather, and leaf-wetness data, while mapped sensor positions let the agent reason across both space and time.
- The Ecowitt gateway posts 62 rows every minute directly into Apache Iceberg; over four million readings occupy 3.22 MB.
- A write-only FTPS drop box receives Reolink JPEGs and indexes their capture time, filename, size, and dimensions into garden.images.
- Full-frame camera images missed caterpillar damage, so the agent now examines full-resolution crops and compares leaves across days for pests, damage, yellowing, and mildew.
- The scheduled garden_morning agent reviews telemetry, images, watering response, historical patterns, and forecasts daily, while retaining a journal, open decisions, and sensor calibration notes.
#lakehouse #llm agents #apache iceberg #iot
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 143, latest observation ; +140 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 76, latest observation ; +76 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hacker News | 50415162 | https://jatorre.github.io/garden/ Read discussion |