Coupon codes that don’t work. Reviews you can’t trust. Prices you can’t compare. The chaos WhatsApp solved for messaging is still sitting there, unsolved, in your shopping cart.
Imagine a maze built for a single afternoon’s entertainment, with a prize of a million dollars waiting at the center for whoever reaches it first. Contestants line up at the entrance, adrenaline running, certain that speed and instinct will decide the winner. Then, before anyone has even broken a sweat, the organizers suddenly announce the winner. You look around, baffled, wondering how anyone could already know the outcome of a race that has just started. The answer is simple and a little humiliating: the winner had a map. Everyone else had a maze to figure out.
That gap — between the person with the map and everyone wandering the corridors blind — is not a metaphor I am reaching for. It is, almost exactly, the condition every shopper lived inside for most of commercial history, and it has a name economists actually use: information asymmetry. The merchant always knew the true cost of the item, the size of the markup, the existence of a better price two streets over. The shopper knew none of it, and had no practical way to find out without spending an entire day walking from shop to shop, comparing, on foot, one item against another.
This is, incidentally, the actual origin of the shopping mall, and the etymology is more literal than most people realize. “Mall” comes from “theMall” — as in you went to visit them all, but it actually was a long, promenade-style walkway, and the shopping mall was conceived, quite deliberately, as a way to put competing merchants under a single roof so a shopper could finally do in one afternoon what used to take a week: walk from storefront to storefront and compare. The mall did not exist because merchants loved competition. It existed because shoppers, given the chance, will always seek out whoever removes the fog first, and eventually someone built a building that did it for them.
Online shopping, in its first wave, was exciting for exactly the same reason the mall was exciting: it looked, for a moment, like it had finally killed information asymmetry for good. Price-comparison engines like the early Priceline promised to show you every price at once. Review sites promised the collective wisdom of thousands of strangers who had already made your mistake so you didn’t have to. For a few years, the fog genuinely lifted.
Then the fog came back, thicker, because now there were a thousand comparison sites instead of one, a thousand review platforms instead of one, and the burden of reconciling all of them landed right back on the shopper’s shoulders. We did not solve information asymmetry. We just gave it more tabs.
The Best Bet Anyone Has Ever Made
A few weeks ago I went looking for something specific: which startup bet, dollar for dollar, produced the highest return any venture capitalist has ever booked. I expected the usual suspects — Google, Facebook, maybe Alibaba. What I found instead was a company almost nobody thinks about when they list the great venture wins, sitting quietly near the top of the table with a return multiple that makes even the legendary bets look modest.
Instagram’s number deserves its own paragraph before we even get to WhatsApp, because the story around it has been sanded smooth by time in a way that erases how insane it actually looked in the moment. In April 2012, Facebook paid roughly $1 billion for Instagram. The team Facebook was buying consisted of somewhere between eight and thirteen people. Not eight departments. Eight to thirteen human beings, with no revenue, no advertising model, no business plan anyone outside the company had seen, running a photo-filter app that had existed for barely eighteen months. The press reaction at the time was not admiration. It was open scorn. Commentators mocked Zuckerberg for what looked, to nearly everyone watching, like a staggering overpayment for a toy — a rounding-error team with a filter app, bought for a sum that dwarfed entire public companies. Baseline Ventures had put in roughly a quarter of a million dollars. On a billion-dollar acquisition, that is a return multiple in the neighborhood of 1,600 times invested capital, generated by a team smaller than most corporate legal departments, in less time than it takes to vest a standard four-year option grant. Everyone who scoffed was, with the benefit of hindsight, watching one of the greatest asymmetric bets in business history and calling it a mistake in real time.
WhatsApp’s number is, if anything, more instructive, because it took slightly longer to prove and involved a business nobody outside a small circle of early adopters had even heard of. In 2011, Sequoia Capital reportedly led an $8 million Series A into that unknown messaging app. Sequoia kept doubling down — a reported further $52 million in 2013, bringing its total commitment to roughly $60 million — and remained WhatsApp’s only institutional investor throughout, a level of concentrated conviction most venture firms are structurally too risk-averse to attempt. In February 2014, Facebook bought WhatsApp for $19 billion in cash and stock, a figure that swelled to roughly $21.8 billion as Facebook’s own stock price rose before the deal closed. Sequoia’s stake, close to twenty percent of the company, turned into a payday north of $3 billion — a return of roughly fifty times its money, on a single, unhedged, five-year bet on two former Yahoo engineers nobody else wanted to fund.
Fifty times is not the interesting number here, not really. The interesting number is what that $19 billion was actually buying: not technology, not patents, not even really a product. It was buying a relationship — the fact that a meaningful fraction of the entire human population had, without any real marketing budget, decided that WhatsApp was where their conversations with the people they loved actually happened.
Why It Had to Be Two Rejects From Yahoo
The story of how WhatsApp got built is more interesting than the valuation, because it explains something the valuation alone cannot: why the company that solved this problem was never going to come from inside the telecom industry it disrupted, and why it was never going to come from a founder chasing a monetization plan either.
Jan Koum grew up in a village outside Kyiv without reliable running water, arrived in California as a teenager on food stamps, and taught himself to program from used textbooks bought secondhand. Brian Acton had spent time at Yahoo and at Apple. The two met at Yahoo, worked there together for roughly a combined twenty years, and left in 2007, disillusioned, by their own later account, with an advertising-driven business model they had come to distrust from the inside. In 2009, both men applied for jobs at Facebook. Both were rejected. Acton, with the dry humor of someone who had nothing left to lose, tweeted about it: “Facebook turned me down. It was a great opportunity to connect with some fantastic people. Looking forward to life’s next adventure.” Two months earlier, Twitter, now X, had turned him down too.
Neither man built WhatsApp because they had identified a trillion-dollar market opportunity in a spreadsheet. Koum built the first version because he wanted a way to put a status update — busy, at the gym, on a call — next to a contact’s name in his phone’s address book, something closer to a glorified away-message than a messaging platform. It became something else almost by accident, and the people who made it something else were not Silicon Valley product managers optimizing a funnel. They were immigrants and their families, scattered across countries, for whom international SMS was not a minor inconvenience but a real and recurring financial cost — a tax on staying close to people you had left behind.
That sentence is worth pausing on, because it is a pattern, not a coincidence. Google did not start as a monetization plan; it started as two Stanford students trying to rank web pages better. In fact, its been reported that they really believed putting ads would go against the mission. Amazon did not start as a logistics empire; it started as an online bookstore run out of a garage, built on the belief that people would eventually trust buying things they hadn’t touched. In every one of these cases, the business model came later, sometimes years later, because the founders were solving a real problem first and trusted that a large enough solved problem would eventually pay for itself. That is not how most startups build today. Most startups now try to prove monetization in the seed pitch deck, because investors ask for it, because “how does this make money” has become the first question rather than the last one — and answering it too early quietly narrows the ambition of what gets built. You cannot build something that reorganizes an entire industry if you are simultaneously trying to prove, on slide fourteen, exactly how you’ll extract a subscription fee and be profitable from it by year two. Keep that thought in mind. It is going to matter again later in this essay, when we get to why most AI shopping assistants today are only a modest improvement on what already exists, rather than the reinvention their marketing promises.
What Made WhatsApp Unbeatable
WhatsApp did not simply arrive with better marketing than SMS. It arrived with three structural advantages that SMS, and the telecoms behind it, could never replicate, because replicating them would have required the telecoms to stop charging for the thing WhatsApp made free.
The first was network effects of an almost perfect kind. WhatsApp did not need you to build a new social graph from scratch, the way a brand-new social network usually does. Your contacts were already sitting in your phone, already the people you actually called and texted. The app simply checked which of your existing contacts were already using it, and for the ones who weren’t, it did something quietly brilliant: it let you invite them directly, and the moment they joined, they were instantly useful to you and you were instantly useful to them, with zero cold-start problem. Every new user made the app more valuable to everyone who already had it, and the invitation mechanism turned every existing user into an unpaid recruiter. SMS had no equivalent mechanism, because SMS was already universal by default — there was no invitation to send, and therefore no compounding advantage to be won.
The second was stickiness, and this is the part product people underappreciate. Once your conversation history, your shared photos, your voice notes, and years of context with a specific person live inside a single thread, leaving that thread has a real cost that a brand-new competitor cannot easily replicate. The medium and the memory become the same object. You do not just lose an app if you switch; you lose an archive.
The third, and this is the one people miss most often, is that WhatsApp’s interface is not really a messaging interface anymore. It is the template. Look at any AI chatbot built in the last three years — the message bubbles, the back-and-forth cadence, the sense of a continuous thread rather than a search query — and you are looking at WhatsApp’s user experience, quietly inherited. That inheritance is not decorative. It happened because a conversation, it turns out, is simply the most natural interface human beings have for interacting with anything that can respond — a person, or, increasingly, a machine. We are not chatting with AI because someone invented a clever new UI paradigm. We are chatting with AI because WhatsApp already proved, at global scale, that conversation was the interface people would default to the moment you gave them a choice.
Ten Dollars, One Phone Call, and a Text That Never Arrived
I have a personal memory that explains, better than any statistic, exactly what SMS actually felt like before something better existed.
I arrived in the UK from Zimbabwe in the early 2000s, a teenager who had just left behind almost everyone I knew. My mum, a single parent, trying her best to provide for us and get us used to being away from home, gave me and my brother a phone specifically so I could call or text back home, and I remember the exact ritual of it: I topped up ten dollars’ worth of credit on T-Mobile, dialed a friend’s number, and told him, breathless, that I was now on the other side of the pond. A few minutes into the call, the line simply hung up — the credit had run out mid-sentence, taking the rest of the conversation with it. I went back and topped up again, and this time, more cautiously, I sent a text message instead. And then I waited. I had no idea if it had actually arrived. You had to specifically activate a delivery-confirmation setting just to know whether the message had reached the tower, let alone the phone, let alone been read. My friend never replied, and to this day I don’t know whether he never got it, or got it and simply didn’t have credit of his own to respond, or got it and just forgot. SMS gave you none of that information. It took your ten dollars, made you guess, and left you holding the silence.
On top of the cost and the uncertainty, SMS was aggressively, almost punitively limited in what it let you actually say. A single message was capped at 160 characters, forcing an entire generation into a strange, compressed shorthand — dropped vowels, abbreviated words, whole sentences squeezed into something that read more like a telegram than a conversation, because going one character over the limit meant either the message got split into two, each one separately billed, or simply refused to send. You could not send a photo without a separate, more expensive service. You could not send a voice note at all. The medium itself was designed around the billing unit, not around what a human being might actually want to say to another human being they missed.
That last detail is the one worth sitting with, because it connects directly to something Mark Zuckerberg has said, consistently, since long before WhatsApp was even on Facebook’s radar: that Facebook’s mission is connecting people. Whatever you think of how that mission has been executed since, the framing itself matters here, because SMS and international voice calls were never priced as though connection was the point. They were priced the way a telegram company prices a telegram — per character, per second, per unit of scarcity — as though the thing being sold was a scarce commercial resource rather than a basic human need to know your family was safe. The telecoms were, in the most literal sense, running a toll booth on connection, and pricing that toll booth exactly the way you’d expect a monopoly to price something nobody could avoid paying for. Facebook, whatever else was true about its motives, looked at the same need and asked a different question: what if staying connected to the people who matter to you was simply free, and you built a business on top of that instead of on top of the friction?
Shopping Never Actually Got Fixed
Online shopping had its own honeymoon period, the way messaging once did, for the same underlying reason: it looked, briefly, like it had killed information asymmetry for good. It has not aged as well as it should have.
Here is what buying something as ordinary as a phone upgrade actually looks like today, if you are the kind of person who takes it half-seriously, which I am. I start on YouTube, watching MKBHD and a handful of other reviewers, because I want to see the camera, the screen, the way it feels in hand, before I trust a single spec sheet. Then I go to GSMArena, because I consider myself enough of a power user to want the actual technical comparison table — chipset, battery capacity, charging speed, display type — laid out side by side against the phone I’m replacing. Then I go back and read more written reviews, because video reviewers and spec-sheet comparisons never quite agree on the things that matter in daily use. By this point I know which phone I want. Now the actual work begins: hunting for the best value. There used to be a simpler answer — eBay, mostly — but now there is a genuinely confusing landscape of secondhand and refurbished marketplaces like Back Market, each with its own grading system and its own warranty terms that don’t map cleanly onto each other. And because I live in the UK, I also have to check whether a mobile carrier deal beats buying outright, knowing that whichever one I pick locks me into a twenty-four-month contract I will spend the next two years mildly regretting regardless of which one I chose.
That is six separate services, four separate trust decisions, and roughly an evening of my life, to buy one phone. And there is a particular absurdity buried inside Amazon’s part of this ritual that deserves calling out on its own: a meaningful number of shoppers use Amazon purely to read the reviews, because Amazon’s review volume and verified-purchase system are genuinely useful, and then go buy the exact same item somewhere else entirely, because the price is better or the retailer is one they trust more for that category. Amazon’s reviews, in other words, are not portable. They live inside Amazon’s walls, usable only as long as you’re willing to also transact inside those walls, which means the most valuable piece of information Amazon holds — the accumulated, verified opinion of millions of buyers — is being used, constantly, as free research for purchases Amazon itself never gets credit for.
Multiply that friction across nearly everything anyone buys, and you get the actual, present-day state of online shopping: not broken, exactly, but exhausting in a way that has become so normal we’ve stopped noticing it as a design failure rather than a fact of life.
Ecommerce’s Telecom Moment
Every major ecommerce platform knows agentic commerce is coming, and every one of them is responding by building their own walled version of it, which is precisely the mistake the telecoms made in the years just before WhatsApp arrived and simply made the entire question of which carrier you used irrelevant.
Amazon rebuilt its own AI shopping assistant, Rufus, and then, in May 2026, quietly folded it into a new product called Alexa for Shopping — merging two assistants that had spent years developing separately and never quite talking to each other, an internal fragmentation problem inside the very company trying to sell you a solution to fragmentation. Shopify has been developing what it calls agentic storefronts, a framework that lets its own merchants plug into AI shopping flows, but one that, by design, works best inside Shopify’s own ecosystem. Google, Stripe, and Anthropic have each published or previewed their own protocols for how an AI agent should talk to a merchant — competing standards, published independently, each betting that theirs becomes the one everyone else has to adopt.
And in the middle of all this maneuvering, Amazon did something that revealed exactly how threatened the incumbents actually feel. In November 2025, Amazon sued Perplexity, seeking to block the Comet browser’s AI agent from shopping on Amazon’s behalf on behalf of its own users, alleging that Comet disguised itself as an ordinary browser to sidestep Amazon’s restrictions on automated tools. In March 2026, a federal judge granted Amazon a preliminary injunction, finding that Comet had accessed Amazon’s site with the user’s permission but without Amazon’s authorization — a genuinely strange legal distinction that only makes sense once you realize what Amazon is actually protecting. Perplexity’s own defense made the economic stakes explicit: agents that shop on a user’s behalf bypass the advertising Amazon shows human shoppers, and advertising, not the sale itself, is an enormous and fast-growing share of Amazon’s actual profit. Amazon was not really defending its customers’ security. It was defending the toll booth.
That is the tell. An incumbent that welcomed agentic commerce would build an open door. An incumbent that fears it builds a lawsuit. Every major ecommerce platform right now is building its own protocol, its own assistant, its own walled agentic layer, competing furiously with every other platform to become the one standard — and in doing so, replicating exactly what the telecom industry did in the years before a messaging app quietly made the entire argument irrelevant.
Why Most AI Shopping Assistants Are Only a 2X
Here is where the earlier point about monetizing too early comes back around, because it explains a real and specific limitation in almost everything currently marketed as an “AI shopping assistant.”
Most of what exists today is, if you are honest about it, search with a chat interface bolted on. You type something roughly like what you would have typed into a search bar, an AI reads a bit more context than a search engine used to, and it hands you a slightly better-organized version of the same links you’d have found yourself, eventually, with more tabs and more patience. That is a real improvement. It is not a reinvention. Call it a 2X: twice as convenient as doing it yourself, not ten times, and certainly not a hundred times, because the underlying interaction — you ask, it searches, you still decide — has not fundamentally changed. It is exactly the shape of an AI shopping assistant built by a team under pressure to demonstrate a monetizable feature by next quarter’s roadmap review, rather than a team given the time to solve the actual, decade-old problem of shopping being exhausting.
A 10X improvement looks different, and it is closer to what network effects and stickiness bought WhatsApp: it becomes ambient rather than a destination. A 10X shopping assistant is one you talk to from your smart speaker while you’re cooking, from your smartwatch on a walk, from your phone in a queue, the same way WhatsApp became the default way you talked to people rather than one app among several you had to remember to open. You don’t visit it. It’s simply there, the way a conversation with a person you trust is simply there, and it places the order without making you open a browser tab at all.
A 100X improvement is a different category of thing entirely, and it is the one worth actually building toward. Imagine telling your shopping assistant you want your living room to feel like a Soho apartment, and instead of returning a list of links, it draws on everything it already knows about your taste, your past purchases, your budget — a kind of second brain built from years of your own shopping behavior — and generates an actual image, or a short video, of your room redesigned in that style, populated with specific, real, shoppable products rendered directly inside the scene. You are not comparing spec sheets anymore. You are looking at the outcome before you’ve spent a cent, and every object inside that generated scene is one tap from arriving at your door. That is the difference between a better search engine and something that actually replaces the entire tedious ritual I described earlier with GSMArena tabs and Back Market comparisons and twenty-four-month contracts: not a faster maze, but no maze at all.
Meet Shopping Claw
This is the setup for a concept I have been developing, provisionally named Shopping Claw: a single agentic layer that sits above Amazon, Shopify, eBay, and every other retail platform simultaneously, rather than inside any one of them.
The mechanics matter here, so it is worth being precise rather than promotional. Today, the retail funnel looks roughly like this: a brand pays for advertising, the advertising captures a sliver of consumer attention, the consumer clicks through to a website or app, and — eventually, if the funnel is well built — a purchase happens. Every stage of that funnel exists to solve a trust and discovery problem that, for a human being armed only with a browser, is genuinely hard to solve alone. Under an agentic model, that funnel compresses violently: a brand optimizes its product data for machine legibility, an AI agent evaluates the option against a standing profile of your preferences, and a purchase happens — with the consumer, for long stretches of the process, simply not present in the loop at all.
Shopping Claw’s wedge is not that it sells anything itself. It is that it makes the comparison layer — the part every platform currently forces you to do manually, tab by tab, coupon code by coupon code — instantaneous and platform-agnostic. It does not care whether the best price for a given item happens to sit on Amazon, a small Shopify merchant, or an eBay listing from a seller with a strong reputation score. It simply finds the answer and executes the purchase, the same way WhatsApp never cared which carrier’s SIM card was in your phone.
The economic argument for why this matters is not abstract. Customer acquisition cost — what a brand spends on search ads, social ads, influencer fees, and affiliate commissions to win a single sale — can eat up twenty to fifty percent of revenue for a typical consumer brand today. That spend exists almost entirely to solve a discovery and trust problem for a human being who cannot efficiently compare hundreds of options on their own. An agent that already knows your budget, your sizing, your past purchases and your standing preferences does not need to be persuaded by a thirty-second video of someone looking aspirational next to a product. It needs machine-readable proof: a durability score, a return rate, a verified price. When the persuasion layer disappears, the cost of acquiring a customer collapses with it — potentially from something like twenty dollars per acquisition down toward the cost of a small API transaction fee, which would represent one of the largest margin expansions retail has seen in a generation.
Ecommerce platforms do not disappear in this future any more than telecom carriers disappeared after WhatsApp. They become what the carriers became: essential, profitable, and no longer the thing anyone actually has a relationship with. The company people trust to know their taste, their budget, and their bad habits — the company that owns the agent, not the inventory — becomes the one with the deepest, stickiest, and most valuable relationship in the entire transaction.
Sequoia did not get rich because WhatsApp built a better telecom. It got rich because WhatsApp made the question of which telecom you used stop mattering. And Sequoia’s $8 million, worth remembering, was not accompanied by a monetization plan either — it was a bet on a team solving a real problem well enough that the money would eventually follow. The company that manages to do the same thing to Amazon, Shopify, and eBay simultaneously will not need to win an ecommerce war fought on any of their terms. It will simply make the terms irrelevant, the same way $8 million once did.
Simba Mudonzvo is a digital marketing consultant, author, and framework builder whose work sits at the intersection of technology, commerce, and consumer behavior. Over an eighteen-year career, he has moved between insurance, big tech marketing, and digital strategy — a path that shaped his instinct for spotting structural shifts before they become obvious. He began his career in Guernsey, spending three years in Generational Aviation & Aerospace at XL Catlin before becoming a Manager in Captive Insurance Management at Willis Towers Watson. In 2014 he returned to London to work in consulting at Marsh & McLennan. From 2015 to 2017, he served as Product Marketing Manager at ASUS, managing a $25 million UK budget and leading the award-winning “Can You Hold Your Laptop Like This?” campaign. He studied Information Systems and Management (BSc) at Birkbeck, University of London. He is the creator of several proprietary strategic frameworks, including Simba’s Five Forces, Simba’s Content Matrix, Internet Presence Optimization (IPO), Customer Ikigai, and Content/Market Fit. He is the author of seven books, including Pied Piper of Digital Marketing, The Gilded Cage, The Emperor’s New Suit, and Marketing 2030: The Future of Marketing When Customers No Longer Shop Alone.
Shopping Claw is one of the core concepts mapped in full inside Everything Becomes Shoppable — the $4,950 report built specifically for venture capitalists hunting for the next asymmetric bet. If you are reading this essay wondering whether you just found the next WhatsApp, you did not find it here. This essay is the two-paragraph pitch. The report is the actual due diligence: the fifty Vibe Shopping scenarios, the Ghost Internet architecture, and the full protocol map of who is realistically positioned to sit above Amazon, Shopify, and every walled garden currently suing its way into irrelevance. Sequoia’s $8 million looked like an ordinary Series A the week they wrote the check. The report costs $4,950. History will not be kind to whoever decided that was the expensive option.