The question arrives from a good place, and it arrives more often now. A client writes to us and says: everybody researches with ChatGPT anyway, so why not help them? Put a small section in the footer, "Ask AI about us", with four buttons, ChatGPT, Gemini, Claude, Perplexity, each one opening the assistant with the question already typed. It looks modern. It looks confident. It says, in effect, we have nothing to hide, go and check us with a machine that has no reason to flatter us.
I understand the instinct completely. For twenty years we have all been taught that third party validation beats self praise, and this feels like the 2026 version of a review badge. But when we look at what actually happens after the click, we do not do it for our clients, and we do not do it for ourselves. Let me explain why, honestly, including the one situation where I would change my mind.
In short: an "Ask AI About Us" link gives away two things at the same moment. It gives away the visitor, who was already on your site and close to acting, and it gives away the story, because the assistant decides which sources to use, which competitors to mention and how your business is described. On top of that, the answer is not fixed, so you cannot check on Monday what a visitor will be shown on Friday. The better investment is the opposite direction: make sure the assistants have accurate material about you, then keep the visitor on your own pages. If the whole idea of AI search is new to you, start with our plain guide to what generative engine optimization means for a business owner.
What people are actually proposing
To be fair to the idea, let us describe it properly. The proposal is a small block, normally in the footer, sometimes on the About page, with a line such as "Do not take our word for it, ask an AI." Each button opens the assistant with a prompt already written, something like "Tell me about Rossi Plumbing in Vancouver and whether they are any good."
The thinking behind it has a certain logic. People already do this research. If they are going to do it anyway, better that they do it from your page, with a question you wrote yourself, than from a blank search box where anything can happen. And there is a marketing bonus: the link says, without saying it, that you are confident.
So the idea is not stupid. It is simply built on an assumption that does not survive contact with the real behaviour of these systems.
Problem one: you are paying to bring people in, then paying them to leave
Start with the simplest objection, the one that has nothing to do with artificial intelligence at all.
Think about what it costs you to get a stranger onto your website. The content, the local visibility work, the profile, the reviews, perhaps the ads. All of that effort has one purpose: to put a person in front of your offer, on a page you control, at a moment when they are considering you. The footer is not a neutral place. The footer is where people arrive when they have finished reading and are deciding what to do next.
And exactly there, at that moment, you put a door to somewhere else. Not to your booking form, not to your phone number, but to a chat window that will occupy the next several minutes of their attention and then, very likely, offer them something else to look at.
One does not need any theory of AI to see the problem. It is the oldest rule of a website: at the point of decision, reduce the exits, do not add one.
Problem two: you are giving away the description of your own business
Here is where it becomes more serious. The moment the visitor leaves for the assistant, you no longer decide what your company is.
The assistant decides which sources it will use, and it does not ask your permission. It may use your website. It may also use an old article from four years ago, a forum thread, a directory page that lists your business with the wrong hours, a comparison post written by a competitor, or a low quality page that exists only to rank. It decides how much weight each of these carries.
It decides the tone. It may describe you as small, or as new, or as "one option among several in the area", which is technically accurate and commercially fatal.
And it decides what comes next. This is the part people forget. These assistants are not built to end the conversation with your name. They are built to be helpful, and being helpful very often means listing alternatives. So the visitor who came to check you can finish that conversation holding a short list of three companies, two of which are not you. You provided the link. On your own site. At your own cost.
We have written before about what happens when you actually ask an assistant to recommend a local business, and the results are a useful cure for optimism.
Problem three: you cannot check a destination that rewrites itself
Now the argument that, in my experience, ends the discussion.
Every other link on your website goes somewhere fixed. If you link to a review profile, you can open it and see what the visitor will see. If you link to a certification body, the page is the page. You can check it once, and check it again when something changes.
An assistant is not a page. The same question, asked twice, can produce two different answers, because the response is generated fresh each time. The model gets updated. The material it retrieves from the web changes. A competitor publishes a comparison next month and it enters the pool of sources. Nothing about this is announced to you.
So this is a link you cannot approve. You can test it on the day you launch it, be pleased, and then quietly ship a different answer to every visitor for the rest of the year without ever knowing. For a website, this is an unusual thing to accept. Nobody would install a testimonial widget that rewrites the testimonial by itself every week.
Problem four: the responsibility does not travel with the answer
There is one more layer, and it is worth stating clearly because it cuts in an uncomfortable direction.
When the wrong information comes from your own chatbot, you own it. This is settled, at least in Canada. In February 2024 the British Columbia Civil Resolution Tribunal decided Moffatt v. Air Canada, a case where the airline's own website chatbot told a passenger something incorrect about bereavement fares. Air Canada argued, remarkably, that the chatbot was a separate entity responsible for its own statements. The tribunal did not accept this, found negligent misrepresentation, and ordered the airline to pay the difference plus interest and fees, a total of C$812.02. The sum is small. The principle is not: a company is responsible for the information on its own website, whether that information sits on a static page or comes out of a bot. The American Bar Association's write up is a good short summary.
Now compare that with the footer link. If a third party assistant tells your customer something wrong about your prices, your service area or your qualifications, the assistant owes them nothing and owes you nothing. There is no counterpart to hold to a standard of care. And yet the customer's disappointment does not float away into the internet. It attaches to your name, because you sent them there.
So the exchange is: no control over the message, no accountability from the messenger, and all of the consequence. This is not a trade a careful business makes.
The one condition that would change my mind
I promised honesty, so here it is. There is a version of this that works, and it is genuinely strong.
If you have actually tested the brand questions, "what is Rossi Plumbing", "is Rossi Plumbing reliable", "Rossi Plumbing versus the alternatives", across all of the major assistants, and the answers come back accurate, current and favourable everywhere, then the link is no longer a gamble. It is evidence. You are showing a prospect that independent systems, working from public material, describe you exactly as you describe yourself. That is a far better proof than a testimonial, because the prospect knows you did not write it.
But notice the order. The link is the reward for having done the work, not a substitute for it. And even then, three conditions apply. Put it on a proof page or a comparison page where a curious buyer goes deliberately, not in the footer where it interrupts the moment of action. Test it again on a schedule, because the answer can change without telling you. And be ready to remove it the day the answer stops being good.
Most companies that ask us about the footer link have never run that test even once. They want the appearance of confidence, before earning it. If you are curious where you actually stand, that is exactly what our free visibility check is for.
What to build instead
If the goal is "I want AI to say good things about my company", then the footer link is the wrong tool, because it works on the wrong end of the problem. Three things work better.
Own the material the assistants read. Build the page you would want an assistant to find: who you are, what you actually sell, where you work, how you compare with the alternatives, what customers ask most often, with real dates and real sources. Write it plainly, in complete sentences, so that a passage can be quoted without being rewritten. This is the whole discipline of getting cited rather than crawled, and the material you publish is the only part of the process you fully control.
Get mentioned in places that are not your own website. Assistants weigh what other sources say about you, which is why brand mentions have become the new backlinks. Directory entries that are accurate and consistent, a real profile on the platforms your industry uses, coverage that names you. This is slow work, and it is the work that actually changes the answer.
Keep the conversation on your site. If people want to ask questions in a chat box, give them one on your own pages, trained on your own content. You will answer correctly, you will not recommend a competitor, and you will get the most valuable side effect of all: a list of the questions you could not answer, which tells you exactly what to write next. Yes, that assistant is your responsibility, as Air Canada learned. That is a feature. Responsibility comes with the ability to fix it.
The point, in the end
The ambition behind "Ask AI About Us" is right. In 2026 a great many buying decisions pass through an assistant before they reach you, and pretending otherwise is not a strategy.
But the response to that shift is not to build a door out of your website. It is to make sure that when someone asks an assistant about your industry, in their own words, on their own time, your business is the one it names, with the right description, and that the page they land on afterwards makes it easy to become a customer.
Being the answer is worth a great deal. Advertising the question is worth nothing.