Do you know where your data lives?
List the systems, spreadsheets and folders that hold the data relevant to your goal. If nobody can say where it is, that is the first thing to fix.
Is it reasonably clean and consistent?
Look for duplicates, missing values and the same thing written in several ways. AI does not remove these problems, it can amplify them.
Does someone own each data set?
Every important data set needs a named owner who can say what it means, how fresh it is and who may use it.
Can the right people access it safely?
Check who can read and change the data today, and whether personal or sensitive information needs extra protection before it is used with AI.
Do you have enough real examples?
Many AI use cases need examples of the right answer, such as past requests with their correct category. A few hundred good examples can be enough to start testing.
Can you measure success?
Decide in advance how you will judge the result: time saved, errors avoided or questions answered correctly. Without a measure, you cannot tell whether the AI helps.
What to do with your answers
If you answered no to several questions, start with a data audit before building anything. If most answers are yes, pick one use case and build a small prototype to test it.