TELL US WHAT
YOU'RE TRYING
TO DO.
Six questions, about four minutes. We ask them because a generic enquiry gets a generic answer — and because half of these questions decide whether the project is worth doing at all. Nothing here commits you to anything.
What is actually going wrong?
Describe it as a problem, not as a technology. "Claims take eleven days and we can't tell why" is more useful than "we want an LLM".
CLOSEST SHAPE — PICK ANYWhere does it live?
The sector sets the rules we have to build inside. The systems tell us how much of the work is integration rather than modelling — usually most of it.
SECTORWhat data exists today?
This is the question that most often kills a project, so it's better asked now than in month three. "We're not sure" is a perfectly real answer.
WHAT YOU HAVEWhat can't move?
Hard constraints are design inputs, not bad news. Tell us the ones you already know about and we'll build the architecture around them instead of discovering them late.
CONSTRAINTS THAT APPLYWho has to agree?
AI projects rarely fail on the model. They fail because the people who own the process were told rather than asked. We'd like to know who they are early.
INVOLVED OR AFFECTEDWhere do we send the answer?
Name and email are all we need. No phone number unless you want a call rather than a written reply.
{{ thanks }}
It goes to an engineer, not a queue. You'll hear back within two working days — and if we think it's the wrong project to build, we'll say that too.