CattleOS

CattleOS combines open outbreak data with herd records to give ranchers local screwworm alerts, smarter inspections, and a practical operating system for 21st-century ranching.

Comment

Inspiration

My family has had a small herd for several years now, and in the light of the New Screwworm, I wanted to build a system for them to easily and intelligently keep track of the their cattle. The hope is to help them stay financially stable and prepared for what’s coming, and to keep track of their inventory as their stock matures.

What it does

CattleOS turns Google Sheets into a cattle-management system for herd, health, breeding, forage, expense, and sales records. Its New World Screwworm Watch uses a ranch ZIP code to display approximate local risk, official-data freshness, prioritized inspection tasks, and relevant TAHC and USDA resources. It also has the potential to link to a GPT account for brief summaries, though authentication is first required (see repo for instructions).

How we built it

I used GPT 5.6 to think of the architecture, had it produce a spec doc, and gave the spec doc to Codex to make the full build. I then verified with Codex 5.6 Sol to troubleshoot and audit the code. Basically, we built CattleOS as a container-bound Google Apps Script application running inside Google Sheets, with deterministic geocoding, geometry, risk, workflow, and validation logic. The system discovers Texas Animal Health Commission ArcGIS layers, provides a safe USDA dashboard fallback, and calls the OpenAI Responses API with strict structured outputs and store: false. From there, we launched into an app with Netlify to make the build easier to engage.

Challenges we ran into

Official public-data sources were not designed as stable application APIs, and the TAHC map used multiple filtered views of the same ArcGIS layer with coded field values. We also had to fit large boundary geometries within Google Sheets cell limits while preserving accuracy, isolate demo data from real ranch records, and maintain a firm safety boundary that never diagnoses animals or determines legal movement permission. It’s still a work in progress and isn’t perfect, but can potentially help cattle ranchers significantly once these problems are fixed.

Accomplishments that we're proud of

I think this app has the potential to help less tech-savvy cattle ranchers keep tabs on their livestock more intelligently. If it helps prevent the spread of New Screwworm, then I will be happy to have help keep the US economy resilient to biological outbreaks in a small way.

What we learned

Reliable agricultural AI begins with trustworthy data boundaries, not merely a powerful model. Deterministic code should calculate locations, zones, distances, and priorities, while GPT-5.6 explains the verified operational picture in plain language. We also learned that clearly communicating uncertainty is a core product feature when livestock health and movement decisions are involved.

What's next for CattleOS

I want to make this available and open source for the public so that ranchers can be as proactive as possible about tending to their cattle. I would like for this to be one of the tools that could help stave off the renewed pest to the southern cattle industry.

Built With

Share this project:

Updates