One page to decide what to build first.
Agents research and cost; you decide with everything on the table, including doing nothing.
Illustrative case. The company and figures are fictitious.
Distribuidora Norteña
Mid-size · 220 people · Monterrey
Which process do we automate with agents in the next 90 days?
CEO, with finance and operations · Gate A
The top 3 options, scored
Priority = (V × F × C) ÷ (K × (1 + T/4))
| Option | V | F | C | K | T (wk) | Priority | Cost and expected return |
|---|---|---|---|---|---|---|---|
| A · B2B quoting agent Builds quotes with prices and stock in minutes. | 5 | 4 | 0.8 | 3 | 6 | 2.13 | MXN 220–280 mil +8–12% quote close rate |
| B · CFDI invoice reconciliation An agent matches invoices, payments and orders; a person approves exceptions. | 4 | 5 | 0.9 | 2 | 4 | 4.50 | MXN 140–180 mil −60% month-end close hours |
| C · HR assistant Answers vacation and payroll questions. | 2 | 4 | 0.6 | 2 | 3 | 1.37 | MXN 90–120 mil −30% HR tickets |
| 0 · Do nothing Month-end close keeps taking 9 days and 3 people. | — | — | MXN 0 today ~MXN 45k a month in hours and errors | ||||
V value · F feasibility · C confidence in the data · K cost to build and run · T weeks until users see it.
Start with B: CFDI invoice reconciliation.
It has the highest priority (4.50), the lowest cost and reaches users first. It also leaves clean data that makes option A more reliable in the next cycle.
Tax data: per-client perimeter and human approval of exceptions.
Runs in parallel with the current process for 4 weeks before replacing it.
At 8 weeks: did close hours drop by at least 40%?