# Building the business case to scale an AI pilot to production

_A successful pilot is not a business case. It is evidence. The case is what turns that evidence into a funded decision, with the return, the cost, and the risk all named._

How to turn a successful AI pilot into a fundable business case to scale. An anonymized founder story on proving value in a live deployment before commercial rollout, plus the return curve to put in front of a board.

# Building the business case to scale an AI pilot to production

A pilot that worked is not the same as a decision to scale. Plenty of good pilots stall right here, because the team assumes the result speaks for itself and shows up to the funding conversation with a demo instead of a case. The business case is a separate deliverable. It takes the pilot's evidence and turns it into projected return, honest cost, and named risk, in the language a budget owner uses.

## Prove it in a real deployment first

An ex-database-engineer turned hardware founder showed me the right sequence. Before committing to a full commercial rollout of an autonomous product, his team ran it in a live deployment for about four months, watching whether it delivered real value to actual customers on their own property. Only once the results were, in his word, promising did they move to commercial deployment. He also priced the ongoing reality honestly, including a subscription to cover continuous software updates and connectivity, because he knew the cost did not stop at launch.

That is the model for a scale decision. The pilot is the four-month live test. The business case is what you build from it: value demonstrated against a baseline, plus a clear-eyed accounting of what running it in production actually costs, including the maintenance and infrastructure that never show up in a demo. A case that only counts the upside is not a case, it is a wish.

## Put a realistic return curve in front of the board

The fastest way to lose a scale decision is to over-promise the timeline. A pilot measures feasibility, not ROI, and the honest curve is roughly 0% during the pilot, 10 to 30% by month 12, and 50 to 150% by month 18. Put that in the case on purpose. A board that is told to expect modest year-one return and a stronger year-two will fund the second year when the first comes in on plan. A board promised instant return will pull funding the moment reality lags the pitch, even when the pilot is doing exactly what it should.

Anchor the whole curve to the baseline you set before deployment, so the projection is an extrapolation from a real number and not an optimistic guess. A return curve with no baseline underneath it is a slide, not a case.

## Name the cost and the risk, not just the win

A credible scale case has three parts, all quantified. The return, on the realistic curve. The full cost of production, including maintenance, infrastructure, and the ongoing operational load, because those are usually larger than the pilot suggested. And the risk, stated plainly: what has to be true for the return to land, and what happens if it does not. Use the pilot's controlled before-and-after as the evidence for the return, so the number you are extrapolating from is defensible. A case that names its own risks is far more fundable than one that pretends there are none.

## How we approach it at Density Labs

Our AI Readiness Assessment is a fixed two week engagement priced at $2,500. When a pilot earns a scale decision, we help build the case: the return projected on a realistic curve from the pilot's measured baseline, the full production cost including maintenance and infrastructure, and the risks named rather than hidden. It is written for the person holding the budget, in their units, so the decision to scale is a business call backed by evidence and not a leap of faith after a good demo.

A pilot proves it can work. The business case proves it is worth doing at scale. Bring both, and the yes is easy.