The shift I see in how people buy: AI as a personal concierge

I think the way people buy services is going to completely change. My belief is that AI will become a personal concierge that understands what someone wants, researches their options, and helps them choose. For an independent service business, that makes explaining who you serve and what you actually do especially important.

By Eric Yim, Protocol Text

What I mean by a personal buying concierge

Looking at my own behavior of how I buy, it's a lot more towards AI. I have AI doing a lot of research for me. That's part of why I think this shift matters.

Today, a person might open several websites, compare what each business says, and work out which questions to ask. I think an AI assistant could take on much of that research with fuller context of what that person is wanting.

That context matters. Someone looking for a service might care about the kind of project a business accepts, the timing, the price, or the way people communicate. Another person looking for the same service could have different priorities.

My forecast is that AI will get better at bringing those details together and making a hyper-targeted recommendation. I'm describing where I think buying is headed, not promising that today's assistants understand every customer or choose perfectly.

A recommendation needs something concrete to compare

If a customer asks an assistant to find the right provider, the assistant needs information that separates one option from another. Broad statements like “great service” leave a lot unanswered.

I'd start with the questions that determine whether an inquiry belongs with your business: What work do you accept? Where do you work? What falls outside your scope? What does someone need to tell you before you can discuss the job?

Those answers give a person something to evaluate. They also give an assistant clearer material to work with if it finds and uses your page.

I also think there's a future where businesses can share their values and consumers can share theirs, with AI acting as a matchmaker. But values need an explanation. If communication matters to you, explain what a customer can expect instead of leaving it at “we care.” Only describe practices you actually follow.

An illustrative example: choosing an office cleaning provider

Imagine an office manager asking an AI assistant: “Help me find a cleaning company for a small office. We need recurring evening visits, and I'd like to know how building access is handled.” This is a hypothetical example, not a Protocol Text client project.

A website that says “quality cleaning for every need” doesn't resolve much of that request. The office manager still needs to establish whether the company takes small commercial spaces, offers recurring work, and can discuss evening access.

Suppose a provider genuinely offers those services. Its page could explain the types of offices it accepts, whether evening visits are available, and what information it needs to assess access arrangements. If it doesn't offer floor restoration, saying so could prevent a mismatched inquiry.

The assistant would then have more specific information to compare with the request. The office manager would still need to confirm availability, scope, and terms directly. A useful recommendation can narrow the search without settling every detail of the purchase.

A practical exercise for your own service page

You don't have to rebuild your website around an AI prediction. I'd begin with one service page and a question a real prospect might ask.

1. Write the whole request. Include the service, location, project type, and any constraint that could change whether you accept it. Use the language customers use when they contact you.
2. Read your page against that request. Mark what it answers, what it leaves unclear, and anything that is outdated. Don't count information that only exists in your head.
3. Add the missing decision details. Explain scope, exclusions, and what you need to know next. Keep qualifications close to the claim they limit.
4. Check the path to an inquiry. Tell the reader how to contact you and what information to provide. Remove requests for details you don't need at that stage.

The aim is a page that helps someone decide whether to have a conversation with you. It doesn't need to anticipate every possible question.

What I wouldn't bet the business on

Clearer pages can't guarantee an AI recommendation. An assistant may miss information, use an outdated source, or misunderstand a qualification. You also can't assume everyone will hand their buying decisions to AI.

I wouldn't abandon marketing that brings useful inquiries just to chase this forecast. I'd make improvements that help people now: accurate service descriptions, clear boundaries, and an explanation of how working with you begins.

There's room for storytelling here, too. I think marketing involves sharing a story and helping people understand what you're trying to do. Keep that voice, but give readers concrete details alongside it. They need enough information to judge whether the story connects with the work they need.

Frequently asked questions

Should I create a separate page written just for AI?

I'd improve the page customers already use first. A separate AI-focused page can create another version to maintain. Keep important service details in readable text, organize them under clear headings, and make sure they agree with the rest of your website.

What if I can't publish a fixed price?

Explain what you need to know before you can quote and which project details affect scope. You can help someone prepare for a pricing conversation without inventing a standard price or publishing a range that doesn't represent your work.

How can I tell whether clearer pages are helping?

Ask new prospects how they found you and what helped them decide to contact you. Notice whether inquiries fit the work you accept and whether the same basic questions keep coming up. Treat those observations as feedback, not proof that a particular edit caused a result.