Why ask before, not after
Once an AI tool is connected to your email, drive or customer records, it can be hard to know what it has already read or copied. A few questions up front save a painful conversation with a client later.
The checklist
- Access: What exactly can it read? What can it change, send or delete?
- Storage: Where is our data stored, and for how long?
- Training: Is our data used to train AI models, theirs or anyone else's?
- People: Who at the vendor can see our data, and when?
- Separation: Is our data kept apart from other customers' data?
- Sending: Can it send anything to anyone without a person approving it?
- Leaving: How do we delete everything, and how do we confirm it is gone?
- Incidents: How and how fast will they tell us if something goes wrong?
Match the answers to your client promises
Read your engagement letters and client contracts. If you have promised confidentiality or limits on sharing, the vendor's answers need to fit those promises.
Start where the risk is lowest
The safest first AI jobs use only public information and send drafts to a person: competitor tracking, new rulings, prospect research. Add access to client data later, one job at a time, when the answers above are clear.
Written by Johnny Hartsfield, coach at Elevate Strategies.
Try one teammate on one job this week. Start free for 30 days.