The nation's fiscal ordinance was clear about what required Council approval above a dollar threshold. The purchasing process was not. Requests moved by email and paper, thresholds were applied from memory, and the accountant reconstructed fund and object coding after the fact from descriptions written by people who had no reason to know the chart of accounts. Every one of those gaps is an audit finding waiting to be written, and federal funding makes the finding expensive.
We built the ordinance into the request itself rather than into training. The dollar threshold now constructs the approval chain: past it, the Council is added, and there is no path that skips them. Fund codes are chosen at the request level and object codes per line item, suggested from the accountant's own chart of accounts and always overridable, with a plain-language explanation above the picker so a staff member choosing a fund understands what they are choosing. Technology purchases route to us instead of guessing at a code.
The output is a coded packet that goes to procurement and the accountant on final approval, rendered fresh from the approved record rather than from whatever was submitted.
We kept the suggestions assistive. The people closest to the money keep the final word, and procurement finalizes. Automating that judgment would have been easier to build and worse to live with.
Requests carry their accounting codes from the moment they are submitted. The threshold is enforced by the system rather than by whoever remembers it. The accountant receives a coded packet instead of a reconstruction project.
I use AI to make good people faster, sharper, and harder to beat, never to replace them. Better, faster, stronger: the goal is a business that grows and hires more humans, not fewer.
No pitch deck. No sales process. Just a straight conversation about what you're facing.