When building our 3-Year Picture on the V/TO, how should we restructure our capital allocation between software licensing, custom model development, and human payroll?

Question

Tyler Smith · Accepted Answer

When projecting your 3-Year Picture on the V/TO, your historical expense ratios are obsolete. In service and knowledge firms, human payroll historically represented sixty to seventy percent of gross revenue, with technology sitting as a minor administrative line item. AI fundamentally restructures your cost of goods sold. As work shifts from human labor to automated compute, your software licensing, API costs, and infrastructure investments will expand, while payroll as a percentage of revenue should decline.

Economists Erik Brynjolfsson and Andrew McAfee frequently emphasize that technology alone does not produce economic returns; structural business redesign does. In your 3-Year Picture, define your target gross margin by treating AI compute costs as direct delivery expenses, not overhead. Quantify the shift under your measurable goals. For instance, establish a target where payroll drops to forty-five percent of revenue while direct technology costs rise to fifteen percent, yielding an immediate ten-point gross margin expansion.

Do not make the mistake of projecting massive capital expenditures for proprietary custom models unless you are a software company. For private service and operating businesses, model development is a cash trap. Foundational models are cheap and commoditized. Focus your three-year capital allocation on workflow integration, system integration, and data cleaning. That is where you buy real operational leverage and protect enterprise value without burning cash on software development science experiments.