Imagine you run a garden design studio. Someone asks an AI assistant what to do with a small yard under a stubborn old maple. The assistant finds your work, likes your approach, and recommends you. The customer replies: “Great. Book a consultation next Thursday.”

Your site says “Call to schedule.” The assistant offers your number. The customer asks for a studio it can book instead.

You won the recommendation and lost the appointment.

That gap has a name: AI transactability. Can an assistant turn a customer’s interest into an actual action with your business? The action might be an order, a booked time, or a qualified request for a quote. If it stops at “Here’s a phone number,” the customer has another decision to make just when they were ready to commit.

The internet is learning a new verb

For years, the main digital challenge was being found. Search sent a person to your website. The person compared, clicked, typed, called, and paid. Now a customer can describe a need, ask for comparisons, and say, “Get me the Thursday slot,” all in one conversation. Your business has to survive the entire sentence.

That takes more than a catchy homepage. What exactly do you offer? Where do you work? What does it cost? What happens if a customer cancels? Are there real openings? A confident recommendation is built from those dull, decisive facts. Action requires one more thing: a working route into the service where you actually take bookings or orders.

The machinery is beginning to arrive. Google’s Universal Commerce Protocol gives eligible retailers and AI surfaces a shared language for shopping and checkout. Stripe’s Link wallet for agents lets a person approve a specific spend request; the agent receives a restricted payment credential rather than the person’s raw card details. Those developments concern supported purchase flows. A service appointment still depends on the assistant and the scheduling provider being able to complete that specific booking.

Think of the path as four separate moments: the assistant finds you, understands why you fit, takes the requested action, and gets confirmation from your business. At each moment, a customer can disappear. The most revealing test is the last one. “The assistant found our booking page” sounds promising; an appointment in your calendar is the result.

What the assistant needs to know

Start with a customer request. A parent wants a Saturday haircut for a ten-year-old, within twenty minutes of home, under a stated budget. An assistant needs the salon’s address, children’s services, prices, hours, and age policy before it can even make a sound choice. Then it needs available times and a route into the salon’s real calendar. At the end, the parent should see the same confirmed appointment the salon sees.

Picture the failure nobody wants: the parent shows up Saturday morning, the child is ready for the chair, and the calendar has no booking. A sentence that says “confirmed” is worthless without the reservation behind it.

The details change for a bakery. A Friday cake order needs the right size, pickup window, price, and allergy information. The final proof is an order number from the bakery’s own checkout. Great photography earns attention; accurate inventory and a completed order earn the sale.

Assistants do not all work the same way. Some research and hand the customer a link; others use a browser or a supported direct connection. Stripe describes these as overlapping paths. That is why “AI-ready” is too vague to be a useful verdict. Name the assistant, the customer action, and the service expected to complete it.

A small-business test worth doing now

Write the request your best customer would make in plain English. Give it a constraint: a date, a location, a budget, a product variation, or a cancellation requirement. Then trace the trip:

  1. Find: Can it identify the right business, offer, and service area?
  2. Choose: Can it resolve the price, timing, conditions, and fit that matter to this customer?
  3. Act: Can it use the booking, checkout, or inquiry service you actually run?
  4. Confirm: Is there an appointment, order, or request in that service, with a confirmation the customer can use?

When a step fails, the repair becomes specific. Your service description may omit the price range people ask about. Your inventory may be current in the store but stale online. Your scheduling provider may offer a compatible connection that has never been turned on. Or it may still require a person to finish the booking. In that case, make the handoff easy and investigate a suitable provider path; a technical badge on the homepage will not put anything in the calendar.

This work should feel familiar to a business owner. A good receptionist needs accurate services, hours, prices, and a working calendar. An assistant needs the same facts in a form it can use, permission to take the next step, and a way to know it succeeded. The appointment still belongs in your calendar. The order still belongs in your store. The money still goes through your payment service.

The garden studio did the hard part: it made someone want its work. The next opportunity is to keep that customer when “I like it” turns into “Book it.” Start with a free AI Visibility Audit to see what your website says about your business and where the path from discovery to action deserves a closer look.