AI Assistant Notice

1. You may be talking to a machine

Some messages from us — on the site, by text and by email — are written and sent by an automated assistant, not a person.

It will tell you so at the start of a conversation, and it will tell you again if you ask. If you ever want to know whether you are talking to a person, ask directly. You will get a straight answer.

We say this because several states require it, but mostly because it is the obviously decent thing to do.

2. What it does

• Answers questions about seawall, dock and deck rules in South Florida
• Looks up what we have published for your city
• Explains what a figure on this site means
• Takes down what you tell us about your property
• Books an inspection slot
• Sends you a guide or a city sheet

3. What it does not do

Without a person involved, the assistant does not and cannot:

• Give you a price or a quote. It can tell you what work has cost in this market, with the disclaimers on our cost page. It cannot price your job, and nothing it says about cost is an offer
• Commit us or any contractor to anything — a date, a scope, a price, a result
• Tell you your structure is safe, unsafe, compliant or non-compliant. Only a licensed professional who inspects it can
• Give legal, engineering, permitting or insurance advice
• Decide anything about you. It does not score you, rank you, approve or deny you, or decide what you are offered or charged

If you ask it for any of the above, it should decline and offer to put you through to a person. If it does not, that is a bug and we want to hear about it.

4. It can be wrong

The assistant can be wrong, incomplete, or confidently wrong — which is worse, because it does not sound uncertain when it is.

It may misread which city you are in. It may quote a rule that has since changed. It may confuse two similar requirements, or miss an exception that applies to you.

Nothing it tells you is a substitute for checking with your building department, and nothing it tells you should be the basis for signing a contract, filing a permit, buying a property or deciding not to repair something.

Where responsibility sits. A person does not review these messages before they go out. That makes the assistant our responsibility, not yours. If it told you something wrong and you relied on it, tell us — we want to fix the underlying data, not argue about it. What we cannot do is accept liability for a decision made on an automated answer rather than on professional advice; section 16 and 17 of the Terms set the limits.

5. What it sees, and where that goes

To answer you, the assistant is given the relevant parts of your conversation and your enquiry record — not our whole database, and not other people's enquiries.

Our AI provider is Anthropic, in the United States. Under our agreement with them, your information is not used to train their models.

We keep conversation transcripts for 24 months, as part of your enquiry record, so we can see what you were told if a question comes up later. They are covered by the Privacy Policy like everything else, and you can ask us to delete them.

6. How to reach a human

Three ways:

• Say so. Type "human", "agent" or "talk to a person". It will hand you over and stop
• Email hello@walldockdeck.com. A person reads that
• Ask us not to use it with you at all. Email us and we will flag your record so that every future message you get from us is written by a person

Asking for a human costs you nothing and does not slow down your request.

7. If it goes wrong

If the assistant said something wrong, pushy, confusing, or just unhelpful, email hello@walldockdeck.com with roughly when it happened and what it said.

We read every one of these. A wrong answer usually means our published data is wrong, which matters far more than one conversation — so you are doing us a favour.