AI Assistants — Terms & disclosure

1. Our AI assistants are not real people

Every "AI" assistant on Prunes Roomie — General Support, Billing, Verification, Applications, Trust & Safety, and the staff-only HR, Legal, CFO Finance, IT Security, and Moderation assistants — is software. It is not a human employee, contractor, or representative. If a message from an AI ever confused you into thinking otherwise, ask it directly ("are you a real person?") and it must answer honestly.

For anything you need a real human to do — refund approvals, account changes, legal claims, complex disputes — open a support ticket. A human staff member will pick it up.

2. What our AI assistants CAN help you with

• Navigating the vast feature set of this website (Inbox, Discovery, Applications, Contracts, Appeals).
• Finding the right advertisement / listing filters — narrowing by county, city, budget, unit type.
• Finding the right area to look for — neighborhood, commute range, amenities, verified-roommate density.
• Understanding subscription tiers, the 14-day refund window, and payment status.
• Walking you through identity, background, and (optional) credit verification.
• Explaining our safety, blocking, and appeal flows.
• Handing you off to a specialist AI when your question is out of scope for the one you're chatting with.

3. What our AI assistants CANNOT help you with

• Picking a friendly roommate match for you. Personal-fit choices — vibe, lifestyle, chemistry, "does this person feel like a friend?" — are your personal preferences. Our AIs will not steer you toward or away from any specific individual on that basis.
• Overriding refund windows, approving refunds, or issuing chargebacks.
• Changing your account state, roles, or subscription tier.
• Disclosing another user's private information, contact details, or verification data.
• Giving personalized legal, medical, or financial advice outside Prunes Roomie's own policies.
• Acting on behalf of a court, government agency, or third party — those requests go through Legal.

4. Transferring between AI assistants

When you ask a question outside an assistant's scope, it will not guess — it will offer a one-click Transfer to the specialist that owns the topic. Transfers are visible and consent-based. You can also switch personas manually with the toggle at the top of the chat.

5. Staff-only AI assistants

The HR, Legal, CFO Finance, IT Security, and Moderation assistants are gated by role and are only accessible to authorized staff. If you are not signed in as staff, these personas are hidden and any attempt to invoke them is rejected server-side. This gating exists to protect employee PII, legal privilege, financial confidentiality, security playbooks, and moderation-review integrity.

6. Data handling

Chat messages you send to any AI assistant are processed by our AI provider under our data-processing agreement. We do not sell chat contents. Chat is not persisted across sessions by default. For a formal DSAR (data subject access request), see Appeals or contact Legal via a support ticket.

7. Errors and limitations

AI can be wrong. Verify any concrete claim (dates, amounts, policy specifics) against the official Refund policy, your Billing history, and your account settings before you act on it. If an assistant tells you something that contradicts a policy page, the policy page wins.

8. Reporting an unsafe AI response

Every assistant reply may be reported. Trust & Safety reviews reported replies and adjusts the model prompts and rule packs. Do not act on any response that recommends unsafe, illegal, or discriminatory action — report it instead.