What shipping in weeks actually costs
A clickable prototype in three weeks is not a discount on engineering. It is a decision to spend the budget on learning first, and it changes what you build.
"Prototype in weeks" sounds like a promise about speed. It is really a promise about sequencing: you find out what to cut before you have paid to build it.
The expensive part is not the code
On most projects the largest line item is work that shipped and was never used. A prototype in front of real users, early, is the cheapest instrument for finding that work — cheaper than research, much cheaper than a rebuild.
What we hold fixed
Speed comes from narrowing scope, not from lowering standards. Three things do not move:
- Architecture. The prototype and the millionth user run on the same foundations. Rewriting the data model at scale is not a saving.
- Correctness where money is involved. Ledgers balance, operations are idempotent, and every mutation is recorded. This is not negotiable.
- Handover. Documented code in your repositories and your cloud accounts, from the first commit.
What we let move
Everything else. Feature breadth, admin polish, edge-case handling, and the second and third integration all wait until something real has told us they are worth the money.