Putting business data into a prompt: a practical risk checklist
A prompt is a shipping label. Before you paste the spreadsheet, know what you are sending, who can see the transit, and what outlives the answer.
A prompt is not a diary. It is a shipping label.
Before you paste the spreadsheet “for context,” separate three residues: the content in the prompt, the transit you do not fully control (provider, logs, retention), and the copies that outlive the answer (tickets, screenshots, shared threads, exports).
Use a checklist, not a gut feeling.
- Whose data is this? Name the subject — customer, employee, household, yourself.
- How identified is it? Names, IDs, account numbers, secrets, credentials.
- How much does the task actually need? Minimum paste beats “dump the file and ask.”
- Where does the request go? Hosted, local, or hybrid — answer in one sentence before you hit send.
- What is retained by default? Mark unknown as promise until a Learn-backed fact exists; do not invent retention numbers.
- Who else can reopen the thread? Shared workspaces and forwarded chats are second hops.
Red lines are short: secrets, credentials, and full customer dumps “just in case.” If the model only needs a pattern, give a redacted example. If it needs a row, give one row. If it needs a policy question, ask without the PII attached.
Sometimes you should not paste at all. Retrieval can fetch what the model needs without shipping the whole store into the box — and that is a different data path (next piece), not a free pass.
This site explains how business data enters AI systems. It does not sell a suite. Named platforms appear later only when a Learn-backed fact is required.