Why an architect building poultry software?#
I came to software from the physical house. A poultry farm is a building that decides a business's profitability for twenty years: its layout, its stocking density and its instrumentation are frozen the day the concrete is poured. What was missing was not another building — it was the layer that tells the owner what is happening inside while there is still something to be done about it.
That background explains two decisions in the system that look odd from outside. First: it starts at the field record rather than the dashboard, because in a house the figure is either taken properly at six in the morning or it does not exist. Second: capture works without signal, because anyone who has stood inside a poultry house knows the signal does not reach and that a system requiring one ends up being a notebook.
What is the thesis?#
In physical industries, day-to-day judgement lives in people who rotate. Software that turns that judgement into a parameter is not selling efficiency: it changes where the asset resides.
A farm manager with twenty years of craft is worth exactly what they are worth, and they are not the problem: the problem is that when they leave, the judgement leaves with them, and nothing written stays behind for somebody else to continue or compare. Management cannot audit a decision that was never recorded, and cannot open a new site without finding another person just like them.
Why from Colombia?#
Because the food chain is a serious investment category, and because the problem is here, where it can be looked at closely.
The mid-sized Latin American producer is too big for Excel and too small for the global integrators' systems. They have poor connectivity on the farm and cannot carry three area managers on payroll. That segment is not a second-rate market: it is a market with different constraints, and different constraints are what produce different products.
What do I actually do?#
I build the system and configure it with each company. This is not software you download and start using: the part that decides needs somebody to declare what that operation is optimising for and what it will not violate, and that happens in conversation with management, not by filling in a form.
I also write up what I find. The articles on the blog come from auditing real flock histories, and they publish the problem with its arithmetic so anyone can check it against their own data. None of them sell anything.