Back Office Crew installs AI assistants (“crew mates”) that draft and perform routine back-office work for our clients. This policy explains the rules they follow.
1. Humans approve first
Every crew launches draft-first. A named person at the client approves outbound messages until a message type is moved to auto-send, in writing, by the client. Sensitive categories (legal, medical, financial, child-related, complaints and anything involving money owed) stay approval-only unless the client agrees otherwise in writing.
2. No professional advice
Crew mates do not give legal, tax, medical or financial advice. They collect information, schedule, draft and route to a qualified human.
3. Disclosure
Where required by law or requested by the client, messages and calls handled by AI identify themselves as AI-assisted. We help clients set disclosure language during setup.
4. Data handling
- Least-privilege access to only the accounts a workflow needs, labeled by client and revocable at any time.
- Client data is used only to perform the client’s work. It is not used to train public AI models.
- We use AI model providers under business terms that restrict retention and training on submitted data.
- Every action is recorded in an activity log the client can see.
5. Accuracy and self-checks
Crew mates confirm facts against the source system, check links and formatting, and flag uncertainty instead of guessing. AI can still make mistakes, which is why approval is the default.
6. Messaging compliance
Texting and calling follow consent and opt-out rules (for example TCPA and carrier requirements). Marketing messages go only to contacts with consent.
7. Shutoff
Clients can pause any crew mate or all of them with one click, or by emailing hello@thebackofficecrew.com.
8. Questions or concerns
Contact hello@thebackofficecrew.com. We review this policy at least twice a year.