The smart farm project is still moving forward, slowly but steadily.
What started as a loose idea — “What if the farm could observe itself and help us make better decisions?” — has started turning into something more concrete. We’ve been calling the system FERN, the Farming Ecosystem & Resource Network, and AURA, the reasoning assistant that will eventually sit on top of it.
The basic goal is simple: build a local-first farm intelligence system that can collect observations, organize knowledge, and help make better decisions without depending entirely on cloud services or corporate platforms.
Right now, the project is still in the foundation stage. That means less flashy magic and more boring-but-important groundwork:
- Building out the local server stack
- Organizing plant and farm data
- Planning sensor networks for water, weather, soil, and security
- Testing Home Assistant integrations
- Exploring LoRa / Meshtastic for low-power farm communication
- Designing an ingestion system for plant, mushroom, insect, and ecosystem knowledge
- Working toward a local AI assistant that can reason over our own farm records
The big principle is that this should not just be “AI slapped onto farming.” The goal is a system that respects local context: our soil, our creek, our weather patterns, our plants, our failures, and our weird little Ozark microclimate.
A useful farm assistant should know that elderberry cuttings behaved differently here than the textbook suggested. It should remember where comfrey thrives, where the wet spots are, what keeps getting eaten, what survives drought, what blooms when the bees are active, and what patterns keep repeating year after year.
That is the long-term vision: not a replacement for walking the land, but a second memory for the farm.
For now, the next major steps are:
- Continue building the plant and ecosystem database.
- Create structured field notes from real farm observations.
- Connect weather, water, and sensor data into one local system.
- Build a safe ingestion loop for adding researched plant traits.
- Test a local assistant that can answer questions using our own records first.
The project is still early, but it is becoming more real with each small piece. The elderberry propagation notes, garden layout records, weather observations, and infrastructure planning are all part of the same bigger system.
The farm is becoming the dataset.
And if we do it right, FERN/AURA becomes the tool that helps us understand it.