Two offers dominate the AI conversation right now, and they look like opposites. One is the free pilot: a vendor offers to show you what AI can do for your operation at no charge. The other is the six-month transformation program: a consultancy offers to remake how your organization works, with phases, a steering committee, and a number long enough to hold a comma.

They look like opposites and they share a spine. Both are shaped around the seller's comfort, and both ask you to commit attention first and evaluate value later. Neither makes the value easy to check: one because nothing is at stake yet, the other because so much is at stake that nobody wants to look at the total again.

There is a third shape, and it is the only kind of first project this studio offers: small, paid, and contained.

What the free pilot really prices

Free is a price, and it tells you how the engagement will be measured. When nobody is paying, the success metric quietly becomes the sale. Your team's time goes in, your operation's data goes in with it, and what comes out at the end is a proposal, because the pilot is the sales call and the roadmap is the product. The invoice shows up at the end wearing the word "partnership."

What the transformation really prices

A six-month program is too big to feel. The value is scheduled to arrive in month seven, after the invoices, and nobody inside the program can say in week two whether it is working. That is a feature of the design: a commitment that large is easier to start than to stop, so it tends to run to the end of its budget either way. Your team absorbs the workshops. Your calendar absorbs the steering committee. And the nagging problem that started the conversation waits politely in the corner until phase three.

What each word buys

Small: one nagging problem, the recurring kind that keeps eating the week. A real answer to one problem is worth more than a promise about all of them.

Paid: paying is what points the accountability at you. When the money flows toward the vendor before the work, the vendor works for the renewal. When the money is attached to a defined deliverable, the vendor works for the outcome.

Contained: a defined scope and a defined end date, set before the work starts. Containment is what makes the experiment easy to read. When the end date arrives, the project has either earned its keep or it has not, and you will know which without a committee.

What contained looks like in practice

The case studies on this site both ran this way. At Oliver's, a Chicago restaurant, the first project was a foot-traffic forecast built from Toast, OpenTable, a weather feed, and a hotel-occupancy tracker: signals the restaurant already generated every day, aimed at the GM's planning question. At the Usher III Initiative, a rare-disease research nonprofit, the first project turned a slow, specialized search into a method the team can rerun as the science moves. In both cases the roadmap conversation, if it happens, comes after the results, which is the correct order.

The decision, priced simply

If a vendor offers to prove themselves for free, ask what the proof will cost you in attention and where the second meeting leads. If a transformation is the only door in, ask to see the first two weeks of it and what those weeks cost. A sharper version of this test, for reading any pitch, is in The chatbot nobody asked for.

Or start smaller than either offer. A paid discovery sprint maps where AI fits your organization, based on the routines and tools you already run, and the results set the pace for everything after.

Prove it on something small. Decide at scale after that.