Andrews Logic & Data Solutions

Real-World Asset Management

Farm_to_Table OS

Pilot Project

Context

Our farm is a living lab for practical AI, and now texts us on WhatsApp.

On our permaculture farm at Learn to Grow Educational Center in Bahrain, we're testing how AI can help manage the physical world, starting with growing food. Our goal is to build a real-world operating system for physical intelligence. Starting with agriculture, but applicable to construction sites, property management, and supply chains.

Demo

Approach

In this demo, you can see how our farm AI is helping us to:

  • Log harvests from a voice note in any language.
  • Keep a searchable "field memory" of everything happening on the farm, building a library of knowledge on regenerative methods to grow food in Bahrain (we now have 2+ years of observational and harvest data!).
  • Manage our task list and suggest tasks based on available time and tools.
  • Turn unstructured observations into structured data and visual reports.
  • Prompt us to make observations that help us track the current state and development of the farm and conduct long-term experiments.
  • Give our farm and food a voice. We can ask our tomatoes a question about their lifecycle and how to best care for them based on our historical observations and research on best practices.

Technologies: Digital Twin • Asset Snapshots & History • Needle-in-Haystack Search • Task Manager • Inventory Manager • WhatsApp Bot • Agent Orchestration

Outcome

This is still in development, but the early results show how much AI can bridge the gap between human observations and actionable insights. Surprisingly, it has meant less time on our phones and computers, more time interacting with and caring for the plants and animals, and more informed decision making.

Before

End-of-day paper checks, missing data, no live view.

After

Notes become data in seconds, most info captured, live pages anyone can check; thousands of observations logged, tons of harvest recorded, hundreds of tasks tracked.

Stack

BuildShipOpenAI & AnthropicElevenLabsSupabasePerplexity

Role & Collaborators

Role: Design the system end-to-end and build it.

Collaborators: Learn to Grow Educational Center farm team and field workers.

Questions we often get (and would love to discuss)

  • Does the efficiency and documentation we gain from AI truly offset the environmental cost of running data centers?
  • Are we anthropomorphizing the farm in a way that risks misunderstanding or misrepresenting its true needs?
  • How do we address the risk of AI hallucinations or incorrect recommendations, given the highly contextual and localized needs of plants and animals in different regions?