Monday Question: Where Does the Document Go?
Before you hand an AI tool a real document, ask this about it:
After the tool finishes, where does that document live?
Not what the privacy page says in general terms. Where the file, the extracted text, and the answer it produced actually sit once the job is done: in storage, in logs, in a cache, in a review queue, in a model provider's retention window.
If nobody can answer in one plain sentence, that's the finding. A vendor who has built for it can tell you quickly. A vendor who hasn't may describe intentions instead. Ask for three specifics: whether the file is stored, whether any log captures its contents, and whether the model provider keeps it past the request. The gap between stateless and zero retention is where those answers tend to go soft, and what proof actually looks like is the follow-up question worth asking.
Try it on the last tool your team put a real document into this month. Write down the answer, then ask the vendor to confirm it in writing.
Getting straight answers from a vendor is usually a short list of questions. Get in touch if you'd like a second set of eyes on yours.