An AI project becomes clearer when the first conversation is about the work. What is taking too long? Where do people repeat the same judgement? What information do they need to make it?
Name the task.
“Use AI” leaves too much undecided. A specific task gives the project boundaries: organise incoming requests, find relevant information, or help a team understand its data. The desired result should be understandable to the people doing the work.
Make the information usable.
Before choosing a model, consider the material it will use. Where does it live? Who can access it? What is missing? A more capable model does not remove the need to understand the information around it.
Keep responsibility clear.
Decide where a person reviews the result, what happens when the system is uncertain, and how the team can correct it. These are product decisions, and they belong in the brief.
Choose a useful measure.
The question is whether the work improves. Define what that means before expanding the project: less repetition, better access to information, or a more consistent process.
Trojan’s AI work spans custom solutions, process automation, data intelligence, and predictive analytics. The task gives those capabilities a purpose.