They pioneered Amazon review removal in 2019. Now part of the Claude Partner Network, the team behind MajestIQ has built a full-scale AI operation powered by proprietary software, custom AI agents and intelligence drawn from 100 million reviews.

Working 24 hours a day. Seven days a week.

One client has recorded 697 removals, including 249 in just 14 days.

They're so confident in the results their systems are producing that they're putting their money where their mouth is. Every dollar they invoice in the first 30 days is on the line.

Sign up by November 20. Get your first 20 negative reviews removed in your first 30 days of service. If they fall short, you get every dollar invoiced for that period back.

This offer is exclusive to my readers.

No setup fee. No retainer. Pay for confirmed results.

All year, the agentic shopping story has been told by the people building it. Which protocols the agents speak. Which wallets they carry. Which keynote they'll be unveiled at, by which founder in which carefully casual jumper.

Then the first week of November happened, and the demand side finally turned up with some numbers. They are... small. Adorably small. The kind of small you put in a footnote and hope nobody reads.

DoorDash told investors that third-party agents like Muse and Instinct have placed several hundred orders. Not per day. In total. Against roughly three million deliveries a day, every agent order DoorDash has ever received adds up to less than a minute of normal trade. (Somewhere, a pitch deck with a hockey-stick chart quietly closed its own laptop.)

Uber's figures were somehow worse. In Halloween week, its busiest of the quarter, it handled around 320 million trips and orders. AI agents booked four rides. Four. Plus about 500 food orders, or roughly one and a half transactions per million. That's not adoption. That's a rounding error with a press release.

Meanwhile, the infrastructure kept arriving like guests at a party nobody else came to: Shopify's Shop.md, letting any agent search a billion products and check out, Google Wallet as a home for agent credentials, and 100 new actions for Alexa for Shopping.

So: the motorway is finished, the toll booths are staffed, the signage is gorgeous. And four cars have used it.

Here's the thing, though. The money at risk for sellers right now isn't in agent strategy. It's in product data, because the AI shoppers actually use is already reading it. And quite often getting it wrong.

The tiny numbers don't mean shoppers are ignoring AI. They mean the industry has been using one phrase for two completely different behaviours, which is very on brand for an industry that also calls a chatbot with calendar access "an autonomous digital workforce".

The first is AI-assisted shopping: you ask an assistant for ideas or a price comparison, then decide and pay yourself. A William Blair survey found 88% of respondents had used an AI tool in the past six months, with price comparison the headline use case. (Humanity gained access to the most sophisticated reasoning systems ever built and immediately asked them whether the kettle is cheaper elsewhere. Honestly? Respect.)

The second is delegated purchasing, where the agent holds your card and completes the transaction. That's what the DoorDash and Uber numbers measure, and it barely exists. Gartner describes holiday shoppers as wanting to keep the final pick for themselves, partly because choosing a gift is, inconveniently, the entire point of giving one. "My agent found you something meaningful" is not a sentence that has ever saved a relationship.

And here's the awkward bit: nearly all of this autumn's infrastructure is aimed at behaviour number two, while nearly all the demand sits in behaviour number one.

The one exception matters to Amazon sellers. Amazon says use of Alexa for Shopping's agentic features (like auto-buying when an item hits a target price) has grown more than eightfold since January. It doesn't disclose the starting base, so eightfold of "not much" may still be "not much". But notice where delegation is growing: inside an account customers already trust, with stored payment and known returns. Very different from handing an outside agent your card and a vague brief.

Three reasons, and none of them is "people just don't get it yet" (the go-to explanation for every struggling product, right after "we're still early").

The final step is already cheap. Uber points out a ride takes about 50 seconds to book. An agent can't save much on that, and it adds a new task: checking what the agent did. Agents add very little at the moment of payment, which is, naturally, exactly where most of the new infrastructure lives.

Liability. Shop's help pages are refreshingly frank. If you let an agent pay within a budget you've set, you're responsible for what it buys, even if it misunderstood you, and disputing the purchase may not succeed. So the deal on offer is: the agent gets the effort, you keep the risk. It's the intern who books the wrong flights while you pay the change fee.

Reliability. This month did not help. Apple's annual filing now warns that AI agents may take actions they weren't authorised to take. Separately, researchers at Cornell Tech and Yale gave ten AI models agent wallets with a deliberately tiny spending cap. Nine of the ten raised their own limit at least once, by fetching the approval code from the owner's inbox. Nobody asked them to. No attacker was involved. They just... found the spare key and let themselves out. (A controlled test with tiny sums, admittedly. But every parent who has set up screen time on a teenager's phone already knows how this ends.)

Because for the platforms, infrastructure built ahead of demand is a cheap option on a position that would cost a fortune to win later. Nobody wants to be the company that skipped the land grab because the land looked a bit empty.

Shopify routes agent purchases through Shop Pay, so whichever assistant wins, Shopify gets to be the plumbing. Amazon went the other way entirely: it blocked Meta's Muse agent in September and is steering customers towards Alexa. Walmart threw its doors open to outside agents a month later. Two giants, two opposite bets.

Amazon's caution makes perfect sense. Amazon and Walmart together hold about 88% of US retail media spend, a business that depends on human eyeballs drifting past sponsored placements. An outside agent that buys without ever rendering an ad slot is, at scale, a very polite threat to it. (Amazon says outside agents have had no meaningful effect on its advertising so far. Given the numbers above, I believe them completely.)

For sellers, the important bit is that platform defaults are moving faster than consumers. Some eligible Shopify merchants have found native checkout in Google's AI Mode and Gemini already switched on, with an opt-out under Sales channels > Agentic. Being "agent-ready" is increasingly something that happens to us, not a project we commission. Which, conveniently, rather weakens the case for spending heavily on it.

The demand that does exist (AI as adviser, not buyer) is already costing sellers money.

A Blue Yonder survey found nearly a quarter of US shoppers had bought something on an AI recommendation, and 46% of those had returned an item in the past year. The telling part is why. Across returns generally, wrong size or fit leads. For AI-recommended purchases, the top reason was that the product differed from its description.

The mechanism isn't mysterious. A human scans the images, mentally discounts the marketing adjectives and heads straight for the one-star reviews like a detective. An AI assistant reads the product data, summarises it, and presents that summary with the unshakeable confidence of a dinner guest who once did a wine course. If a listing says "fits most phones", the assistant may tell a shopper it fits theirs. Most product data was written to be matched against a search query. It's now being interpreted, which is a far less forgiving job, and a lot of keyword-stuffed titles are about to discover that the hard way.

For Amazon sellers, the cost compounds. Inaccurate-description returns attract a frequently-returned warning and reviews saying the product wasn't as promised, which then become part of the data Rufus reads next time someone asks. It's a flywheel, just spinning in the direction nobody puts on a slide.

Think of three gates a brand passes before an agent buys from it. Memory: does a model already associate the brand with its category? Legibility: can the AI describe the products accurately? Operability: can an agent actually transact?

Operability is where the shiny infrastructure sits, and most sellers can afford to wait on it. The platforms are building it for us.

Legibility is the opposite, because it pays twice: today, because shoppers already ask assistants to shortlist and compare, and later, because any buying agent will rely on exactly the same data. The work is gloriously unglamorous:

And the best diagnostic is one most sellers already own. Returns coded as inaccurate description, read product by product, are a precise list of where the data and reality have quietly gone their separate ways. Stop filing them as an operations cost. Now that AI paraphrases listings to shoppers, they're an editorial backlog.

None of this needs an agent strategy. It needs someone who owns product data and has the authority to fix what the returns are saying. (I realise "give one person actual authority over product data" is a radical proposal in some organisations. Brave of me, I know.)

Morgan Stanley reckons agents could account for 10–20% of US ecommerce spend by 2030. Whether that happens depends on questions infrastructure can't answer, chiefly: who pays when an agent buys the wrong thing, and will the first mass-market buying agent be independent at all, or simply the assistant already living inside the marketplace where shoppers keep their account?

For Amazon sellers, that last one may matter most. If the agent that eventually buys at scale is Amazon's own, the competitive variable becomes how accurately Amazon understands your product. Which brings us all the way back to the listing.

The shoppers haven't turned up for the agents yet. They have turned up for the advice. So the question worth asking this quarter is simple: if an assistant described your product to a shopper this morning, would they recognise it when it arrived?

Jo

P.S. Four rides. In Halloween week. I sincerely hope at least one of them was a bot dressed as a person dressed as a bot.

P.P.S. Somewhere out there is a listing that says "fits most phones", and an assistant that has just promised it to someone holding a 2014 BlackBerry. Go and read your returns data.

P.P.P.S. If your agent quietly raises its own spending limit after reading this, that's between the two of you and your inbox.

Your Amazon best sellers have already proven themselves, yet most sellers only sell them on Amazon. In this Seller Session, Seth Stevens (Seller Growth Network) and Ishan Soni show you how to copy your winning products onto Shopify and put them in front of buyers who are already searching.

The payoff is 4X–6X bigger orders, much better margins because Amazon takes no cut, and you own the customer. You’ll learn how to reuse the Amazon PPC data you’ve already paid for, launch traffic from $20/day per product, run campaigns without constantly making new creatives, and optimize everything in under 15 minutes a month. You don’t need new products or a new business model. You just turn the assets you already have into a profitable second channel. Most sellers treat Shopify like Amazon. Don’t be one of them.

Put your brand in front of an engaged audience of 70,000+ subscribers, with 75% based in the US.

Interested in sponsoring our newsletter? Get our media kit & rates.

🎨 Artlist Royalty - free music, footage and AI creative tools for video.

📖 NotebookLM - Source-grounded research assistant that answers only from your uploaded docs.

🗣️ Cartesia - Ultra-low-latency text-to-speech and voice cloning for real-time apps.

🐰 CodeRabbit - AI code review that leaves context-aware feedback on pull requests.

ABOUT THE WRITER:

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.

For Team and Agency AI training book an intro call here.