27 July 2026 · Jacques Theron · Markdown source

Day one, in public — what GroundYield is and is not

This project is one day old as a public stack: domain, GitHub identity, website, and open documents landed in the same window. That belongs at the top — not in a footnote.

GroundYield is an attempt at a modular smallholder farm pilot near Vilanculos, Inhambane Province, Mozambique, with costs, yields, and failures published openly. Operator: Jacques Theron. Self-funded. No legal entity yet. No local partner named on the accountability page, because none has agreed in public.

One line: nothing has been measured on a pilot plot. Zero units installed. Zero local baselines. Zero written supplier quotes. Trip dates open.

Scars first (not a blank CV)

I am not a tourist in Vilanculos. I previously farmed at scale here (own operation ~3,500 ha; larger partnership/Toro-era context), then left after COVID-era costs and bad choices — operator-disclosed loss order ~ZAR 10 million. I do not claim those hectares today. Day job: HSE shutdowns, Anabeeb, KSA. Return focus: Rumbacaca — listen first.

Full story: LESSONS_MZ.md · Risks: RISK_AND_GAPS.md. GroundYield is the opposite bet: small modular units, open costs, public failures.

What exists

What does not exist

Why publish before the trip

Most projects write measurement rules after the first good photo. We are trying the opposite: anti-fraud and data rules first, so inventing wins has something public to collide with.

Full policy: INTEGRITY.md

Adversarial review (what it got right)

Day-one critique: docs ahead of ground contact; self-referential trust; targets before feasibility; early “shipped” claims briefly outran live HTTP 200s; agronomy drafts must not go live without review; households must not be enrolled into silent abandonment.

Those points were substantially right. Deploy is healthy now (production equals main). Strategy risk remains: more markdown does not create a pilot.

AI disclosure

Much of the documentation and site work was AI-assisted (including tools such as Grok), under my direction. I am accountable for what is published. If the prose looks machine-helped, it often was. Concealment would be worse. The test is honest stage and numbers — and public corrections when we fail.

What we are asking for

Not money in this post.

team@groundyield.org · GitHub issues

Start here

LinkWhy
HomeStatus + start paths
White paperLiving thesis
Docs libraryFull map
Unit registryCount = 0
LESSONS_MZPrior scars
X thread copyReady to post
OUTREACH.mdHow we promote

If this only produces documents, we failed — and we should say so. If it produces data, that data will be public either way. Pressure-test us.

← Home · Docs · White paper · Roadmap · Português