Most software in this trade arrives finished, with a price and a testimonial from someone you can't call. This one is being assembled where you can watch it, and the mistakes stay in the log.
17 September 2026
Came off noindex. Sitemap, robots, and an IndexNow key so Bing and Yandex get told directly rather than waiting to be crawled. The argument for indexing a half-built product site is that the half that exists is honest, and the parts that aren't built say so.
17 September 2026
The seven structural defects, the one tree-length rule, and the tree care year. Stock photography of a tree says nothing, and this trade can tell.
17 September 2026
A registry check refused the project: serves treepro.io with NO git link. Nothing can reproducibly change this site. It was right. The site had only ever been pushed by direct file upload, so there was no path from the repository to what was live.
Now linked. A push to main rebuilds it, which is the only way anything automated can safely touch the site.
17 September 2026
Multi-tenant or repo-per-tenant, and which stack. Both had been parked as though someone else needed to rule on them. They were engineering calls, they got made, and they are written down as decision records with the reasoning attached.
A tenant is a database row. Repo-per-tenant freezes every customer at the version they signed up on, which is fatal for something whose whole proposition is that it keeps getting better.
17 September 2026
First draft framed TreePro as an agency service with a vertical focus. That was the wrong shape and it took a blunt correction to see it. Rewritten as a product: one codebase, many tree companies, with the arboriculture encoded rather than delivered by hand.
17 September 2026
The first research pass searched for "tree" and came back full of binary search trees and network spanning-tree protocol. The second was scoped properly and still thin, because a 30-day window on a trade this size does not hold enough operator conversation.
What actually produced the product wedge was software review sites. Tree companies describing, in detail, what the tools they pay for do to them. That is where the proposal-delivery problem came from, and it is why this product treats a delivered proposal as a feature rather than an assumption.
Everyone selling to this trade arrives with a polished site and no evidence. Operators have learned to assume the polish is the product.
A log with the mistakes in it is checkable in a way a testimonial is not. If something here turns out to be wrong, it stays and gets a correction underneath rather than being quietly edited out.
When there are customers and real numbers, they will appear here with names attached and permission on file. Until then this page is what we have that is actually true.
The first crews on this will be named here, if they want to be, and quoted accurately if they don't.