Before I scope anything, I talk to the people doing the work today and watch how they do it. The problem they describe is usually different from the one in the project request, and the difference is where most of the value is. Writing an accurate problem statement is the slowest part of a project, and I do not skip it.
AI makes it cheap to build the wrong thing quickly. So before any code is written, we agree in writing on what a finished, working system does, and I build toward that definition until it passes.
I did this work for years before AI coding tools existed, and the steps have not changed. I still write requirements, still draw mockups, and still read generated code against the same standards I applied to hand-written code. The tools made each step faster. They did not remove any.
Every engagement ends with documentation and training sessions. Whether I host the product or you do, your team gets the source code, the architecture docs, and enough walkthrough time to extend or replace any part of it without hiring me back.
When an organization needs a data system built, it usually has three options: hire a consultancy, hire a full-time employee, or bring in a contractor. Each one fails in a specific way.
A consultancy staffs the project with a team, bills for the team, and often keeps the code and the know-how when the engagement ends. A full-time hire takes months to recruit and stays on payroll after the project is over. A contractor fills a seat on your team but is rarely accountable for whether the finished system works.
JMS Analytics is one person who joins your team for six to twelve weeks, works out the real problem with the people who have it, builds the system with AI tooling, and hands it over with documentation and tests. When the engagement ends, your team owns the system and I leave.
