Map the process before automating it, seven states wide
The same method cut my estimating time and standardized a cannabis operator's procedures across seven states.
Estimating across my own companies once ran about six weeks per job. On a separate track, a multi-state cannabis operator wanted one set of standard operating procedures covering seven states, each with its own regulatory weight.
For neither problem did I buy software off the shelf. I wrote proprietary automation covering plan scanning, takeoffs, and cost correlation by market. Then I used the same approach on the cannabis client's procedures: map the process first, automate second.
My view of AI is that it is engineering applied to a workflow I already understand, and no shortcut around understanding the business. If you model the process first, what gets automated stands on metrics you already trust. That is why one method moved cleanly between two unrelated industries.
Estimating that took six weeks now takes about six hours inside my company. The cannabis client received standardized procedures for all seven states, with nobody reconciling each state's rules by hand.
Automation pays only when you understand the process well enough to know what deserves encoding. Neither client received a black box from me. Each got automation sitting on a process I had already proven I could run manually.
Story details
| Type | Insider truth |
|---|---|
| Year | 2026 |
| Firm telling | on the Common Ground wiki |