No. Legal and financial questions get declined and handed to a person as a standing rule, the same way med spas and gyms get their own tuned boundaries.
No. The assistant is built to hand off legal or financial questions to a person rather than answer them — that boundary is a standing rule in how it's built, not a judgment call it makes on the fly. It still checks every other answer against the business's published pages, and a separate deterministic check runs in code for crisis and hazard situations before any model even sees the message.
Some fields get boundaries tuned to what matters there. Med spas get a no-medical-advice line with handoff to a clinician, and gyms route injury questions to staff and to 911. Legal and financial questions get the same kind of boundary — the assistant declines and hands off rather than answering — as a standing rule in how it's built.
That boundary is a standing rule in how the assistant is built, and every answer is separately checked against the business's published pages before it ships; separately, a deterministic crisis-and-hazard check runs in code before any model sees a message. The two checks are different things. The published-content check keeps ordinary answers accurate. The code-level check exists for a narrower list — things like crisis or acute distress, gas or fire, electrical hazards, flooding, hazardous DIY situations, and is-it-safe or permit questions.
It doesn't tell a visitor what to do with their money, how to interpret a contract, or how a law applies to their situation. When a question crosses into that territory, it stops and hands the visitor to a person instead.
A visitor to a financial advisory firm's site asks: "Should I move my 401k into an index fund?"
The assistant answers: "That's a decision I can't weigh in on — let me get you to someone at the firm who can. Can I get your name and the best way to reach you?"
A model deciding, case by case, whether a legal or financial question is safe to answer would be unreliable — the same question could get answered one day and declined the next. Building the decline-and-handoff behavior in as a standing rule removes that guesswork. The assistant isn't weighing whether it feels confident enough to answer; it's following a boundary set in how it's built, every time.
That code-level check is separate from the field-tuning above and runs before any model sees the message at all. It's reserved for a specific, narrow list of situations rather than every sensitive topic a visitor might raise — which is why a legal or financial question is handled through the built-in standing-rule boundary described above, not through that code-level gate.
No. It's built to hand those off rather than guess, and it never invents an answer it can't verify.
No. Boundaries are tuned to the field — med spas and gyms are two examples, and legal and financial questions follow the same declines-and-hands-off approach.
The business owner. The assistant captures the visitor's name and contact, with their permission, and sends it to the owner's inbox — even at midnight. Try the free private preview — built in under 24 hours, no card — at askmy.business.
Give us your website and we build your private assistant — free, usually within a day, no card. Nothing goes live without your approval.
Build my free previewUpdated 2026-09-01 · AskMyBusiness by Vision Genesis · Knoxville, TN