About

We Build terminal_

A platform that turns natural language into autonomous action — deployed on your infrastructure, running on your terms, owned by you.

The Founder

R. Desmond (TD)

Founder & CTO · Desmond Digital Services

24 years consulting and farming across Oregon and Central Texas. Built enterprise systems for industries ranging from agriculture to aerospace. Now focused full-time on making autonomous AI accessible to the people who need it most — operators in the field, not just engineers in the office.

24 Years Combined Consulting + Farming
2015 First Farm Automation Project
Jun 2024 Open Interpreter Live Demo

Live Demo — June 2024

40-minute live demonstration to scientists and developers from major tech companies. Showed autonomous multi-step AI workflows running on iOS — months before Apple's WWDC 2024 announcement of similar capabilities.

What Was Demonstrated

  • Autonomous multi-step workflows (1 prompt → 6+ AI steps)
  • Persistent memory with daily compression
  • Cross-device orchestration (iPhone ↔ Mac ↔ iPad ↔ Raspberry Pi)
  • Artifact generation (HTML dashboards, spreadsheets, visualizations)
  • 60+ iOS Shortcuts as AI-executable tools
  • Specialist AI personas (breeder, agronomist, market analyst)
What We Build

Desmond Digital Services builds terminal_ — an autonomous AI agent platform that works in any industry. FarmFriend is the first vertical: agriculture.

terminal_ Platform

The core IP: dual orchestration AI runtime that routes tasks between parallel and hierarchical processing. Works with OpenAI, Claude, Groq, or local models. State persistence across 8+ hour autonomous sessions.

See the Architecture →

Future Verticals

terminal_ is industry-agnostic. The same architecture that runs FarmFriend can be adapted to healthcare, logistics, manufacturing, or any domain. FarmFriend proves it works. The platform enables it everywhere.

Open Source

A persistent AI coworker that lives on your infrastructure — reading files, running commands, automating tasks, coordinating devices. Built with TypeScript, deployed on edge hardware, controlled by voice or text.

TypeScript ★ 1 Fork 1

FFT_model_tune

Fine-tuning

LoRA fine-tuning of Trinity and Qwen3.5 models for specialized agricultural AI. Training on domain-specific farm data to improve response quality for crop, livestock, and equipment contexts.

Python

agent-skills-farming

View on GitHub

Specialized agent skills for farming operations: irrigation scheduling, pest monitoring, livestock tracking, and yield forecasting. Built on the terminal_ skill architecture.

TypeScript
How We Work
01

Open Source First

Our software is MIT licensed. Your data stays on your farm. No subscriptions that hold you hostage. No vendor lock-in.

02

You Own It

FFT_nano runs on your hardware, on your network. The AI works for you, not the other way around.

03

Farmer-Tested

We run FarmFriend on our own operation. Every feature is battle-tested before it ships. If it doesn't work in the field, it doesn't ship.

04

Composable Architecture

terminal_ is a platform, not a product. Build on it, extend it, adapt it to your industry. FarmFriend is just the first example.

See terminal_ In Action

FarmFriend demonstrates what the platform can do in agriculture. The same architecture works in any industry.