Your Next Customer Might Be an AI Agent
Picture someone looking for exactly what you sell. But they never visit your website. They tell an assistant: "find me this cheaper, delivered by Friday, with decent reviews." The assistant checks a dozen stores, compares them and drops the best option into a cart.You'll never know you were in that contest. Or that you lost it.This isn't a forecast. In mid-October, this is how people in the US started shopping. And looking at the news from that week, one thing is clear: the biggest tech companies are building a layer of AI intermediaries between businesses and their customers, all at once.
What happened
Google: a cart that compares on the shopper's behalf. On October 15, Google rolled out Universal Cart across Search and Gemini in the US. It's one cart for products from different stores. It tracks price drops, flags when an item is back in stock and surfaces promo codes. Shoppers can pay with Google Pay right there or hand the cart over to the store's own site. Early partners include Nike, Sephora, Target, Walmart, Wayfair and Shopify merchants, with Canada, Australia and the UK next.One detail matters for every business: the store stays the merchant of record. But searching, comparing and deciding no longer happen on the store's website. They happen inside Google. And Universal Cart only sees products whose data the store has sent to Google Merchant Center: price, availability, specs.Amazon: an assistant that remembers what you promised. The same day, Amazon's Bee released the Bee Frontier wristband at $74.99 and the Bee Eclipse at $99.99, which attaches to a phone. Both listen to conversations and turn them into summaries, reminders and to-dos.It sounds like a gadget for forgetful people. Now look at it from the business side: your customer now has a written summary of every call with your sales rep. What was promised, by when, at what price. Their assistant will remind them. Who reminds your rep?Apple: an assistant at the center of the home. On October 13, Apple showed HomeView, a home display built around its new Siri. Apple had held it back since 2024 precisely because the device made no sense without a capable assistant. One more place where people ask out loud instead of searching with their eyes.
Why this isn't a one-off
The rest of that week's news explains why there will be more agents and why they'll keep getting cheaper.Models are converging. Meta's new Muse Wonder Full scored 56 on the Artificial Analysis index, against 58 for the leader. A good "brain" for an agent is no longer rare.Running AI is getting cheaper. Microsoft says its new Maia 300 chip delivers 40% more tokens per megawatt. The AI cloud Modal grew from $300M to over $1B in annual revenue in five months, because demand for running AI is enormous. According to Stanford, the cost of a GPT-3.5-level query fell from $20 to $0.07 per million tokens in under two years.And agents already hold conversations by the millions. Australian startup Apate says its AI agents, posing as scam victims, have had 2.5 million conversations and wasted 17 years of scammers' time. If an agent can keep a fraudster on the line for hours, comparing five offers on a shopper's behalf is easy.The conclusion is simple: cheap, capable agents will show up wherever a person makes a choice. And one of the options they're choosing between is you.
What an agent sees when it "looks" at your business
A person can be won over by a banner, a great photo or a charming sales rep. An agent can't. It reads data. And if the data isn't there, you don't exist for it.Here's what an agent looks for first:Price. A real one, on the page. "DM for price" reads as "no price" to an agent, and it moves on to someone who lists one.Availability or open slots. Is it in stock, when can it ship, is there a free slot tomorrow? If only your front desk knows, the agent doesn't.Description in text. Specs, terms, what's included. Not in an image or a story, but in text that can be read.Reviews. Recent ones, on open platforms: Google, Clutch, marketplaces.Policies. Delivery, returns, cancellations. Agents compare those too.Response speed. When an agent does send an inquiry, whoever replies in minutes wins, not whoever replies the next morning.
This isn't only about online stores
It's easy to think "I'm not in e-commerce, this doesn't apply to me." But the logic is the same for any business where customers choose between options.A clinic where open slots live in the receptionist's head. A logistics company that calculates every quote by hand. A studio where you can only book through DMs. For a person, that's "a bit inconvenient". For an agent, it's "no data, next option".We've seen this from the other side. For one clinic, we built a system where AI replies to patients instantly, matches them with the right doctor and shows open slots in real time. The result: 45% more appointments with the same team. In effect, we made the clinic's schedule readable by a machine. That's exactly what it will take for your customer's agent to find you.
Check yourself in 10 minutes
The simplest test you can run today:Open ChatGPT, Gemini or another assistant and ask it to recommend a company in your niche, in your city or country. Are you in the answer? What does it say about you?Ask it directly about your company: prices, terms, how to book. Is the answer right? Where did it come from?Open your website and look at it like a machine would: are price, availability, terms and contacts there as text, without clicking into a messenger?Results vary by assistant and change over time. But they honestly show the main thing: whether you exist for an AI intermediary.
What to do about it
You don't need to rebuild your business for robots. You need your basic information to be available without a human in the loop: prices, availability, schedule, terms, reviews. And you need to answer inquiries fast, whether they come from a person or an agent.A nice side effect: real customers like all of this too. People don't enjoy sending a DM just to find out a price either.If you'd like to see how an AI agent sees your business, we run a free AI audit. In 30 minutes we find where your data is "hiding" in people and messengers, and what to automate first.