From an enquiry to a prepared conversation.
Studio OS is the system I am building for my own practice. It brings an enquiry, a short reading of the problem and questions for the first conversation together in one private review page.
Status: a working local prototype. The public enquiry form is not open yet. Tests use invented enquiries, and no client results are claimed here.
The problem
A useful first conversation needs context. Preparing that context should not give an AI permission to make promises or send a reply.
The design
Code picks out what is unclear and prepares the questions, with volume, cost and approvals first. A model can restate the problem and sketch the current process. If that reading fails, the questions still appear and the page says no model read it.
The pre-read sits beside the enquiry. A person decides what happens next. Nothing in this flow sends a reply automatically.
My role
System design, direction of the AI build, testing and operating decisions. I use tests that fail before a change and check the real flow in a browser. The code has also been deliberately broken to check whether those tests notice.
What I am still checking
Local tests do not prove live hosting, provider policies or a better conversation. The form's privacy and operating decisions must be settled before opening it to real enquiries.