So Daniel sent us this one — he's deep in apartment setup mode with Hannah, ordering a ton of small parts from AliExpress, and he's running into the same friction everyone hits. You're mid-project, you realize you need a two-dollar keystone jack. Can't justify four bucks in shipping for it. So you wait, meaning to bundle it with other items. And you forget. And the project stalls for a week. The core question is, what would it take to build a system where anyone in the household adds items to a shared cart, and when it hits a threshold, the order just places itself? And how close are we to that actually working?
The gap between what Daniel wants and what exists is surprisingly wide. And I think the first thing to establish is why cross-device cart sync doesn't work reliably, because that's the foundational frustration here. AliExpress and most e-commerce platforms don't actually sync your cart in the way people assume. What they do is store your cart server-side, tied to your account, and then use session cookies to maintain continuity within a browsing session. When you switch from your phone to your laptop, the cart data is supposed to follow you because you're logged in. But it breaks constantly.
And the reason it breaks is that the session management isn't really designed for cross-device continuity. It's designed for single-device persistence, and cross-device is a side effect they sort of hope works.
The cart sync relies on browser cookies that are tied to your login session. When those cookies expire, or when you use incognito mode, or when the platform's sync logic has a race condition between the mobile app and the web client, the cart state gets inconsistent. I've seen forum threads going back years where users report items disappearing when switching devices. AliExpress hasn't fixed it because, fundamentally, the architecture wasn't built for this use case. It was built for you to browse and buy on one device.
So the "sync" is more of a hope than a feature.
It's an aspiration with a loading spinner. And that's just the first pain point Daniel identified. The second is that there's no shared cart across users. Even if the sync worked perfectly, his cart and Hannah's cart are separate entities. They're shipping to the same address, they both want to hit that ten-dollar free shipping threshold, but there's no mechanism to pool items across accounts.
And the third pain point is the mental overhead. You're keeping a running list in your head of the four or five small items you need, trying to remember not to pull the trigger until you've accumulated enough, and meanwhile you're doing actual projects where you need the parts.
This is exactly what Daniel's getting at. The system he's describing — shared cart across users, auto-placement at a threshold — that's not some sci-fi fantasy. That's how enterprise procurement works. It's just completely absent from consumer e-commerce.
Let's talk about that enterprise comparison, because I think it's illuminating. What does Coupa or SAP Ariba do that AliExpress doesn't?
In a corporate procurement system like Coupa, you have something called an auto-approve workflow. An employee adds items to a requisition cart. If the total is under, say, five hundred dollars, the system auto-approves it and sends the purchase order to the supplier without any human in the loop. Above that threshold, it routes to a manager. The key is that the rules are configurable, the cart is shared across the organization, and the checkout is automated. It's exactly the pattern Daniel wants, just in a business context.
So the capability exists. The question is why it hasn't migrated to the consumer side.
Three reasons. First, consumer platforms don't expose APIs for cart manipulation or checkout triggering. Enterprise procurement works because the buyer and supplier have negotiated integration points. AliExpress has no incentive to let a third party programmatically add items to carts and trigger checkouts. Second, liability. Auto-placing orders means auto-charging credit cards. If the system orders the wrong thing, or orders too much, or orders at a bad price, who eats that? The platform doesn't want to be in the middle of that dispute.
And the third reason is the one that stings — they don't want you to automate this. The friction is the business model.
Right. Platforms want you to browse. The more time you spend on AliExpress, the more likely you are to see something else and add it to your cart. Auto-ordering eliminates the browsing session entirely. You add the keystone jack, the system places the order at threshold, and you never see the "recommended for you" section. That's a direct threat to their revenue per session.
So the feature gap isn't an oversight. It's a deliberate choice dressed as a technical limitation.
And that's important context before we look at what actually exists to bridge this gap. Because everything we're about to discuss is a workaround against platforms that actively don't want to be worked around.
So let's start with the closest thing to a solution that exists today. You found something called Share-A-Cart?
Share-A-Cart is a browser extension that does one specific thing well. You're on AliExpress, or Amazon, or Walmart — it supports a bunch of major platforms — and you've built up a cart. You click the extension, it generates a shareable link that encodes your exact cart contents. You send that link to someone, they click it, and it imports those items into their cart on the same platform.
So not quite the shared persistent cart Daniel's describing, but it solves the cross-user transfer problem.
It solves the cross-user transfer problem manually. The workflow would be: Hannah adds three items to her cart, generates a Share-A-Cart link, sends it to Daniel via WhatsApp. Daniel imports those items into his cart, adds his own items, and when the combined total hits ten dollars, he places the order. It works. It's just not automatic.
And it doesn't solve the threshold monitoring problem. You still have to remember to check the total.
You do. And you have to remember to generate and send the link each time. There's no merge, no auto-update. If Hannah adds another item after sending the link, Daniel's cart doesn't know about it until she generates a new link. It's a snapshot, not a sync.
So it's the digital equivalent of texting someone your grocery list and hoping they combine it with theirs before checkout.
That's exactly what it is. But it's the best off-the-shelf tool we found for the cross-user piece. Now, for the threshold monitoring piece, there's nothing consumer-facing that auto-triggers checkout. But there are some partial automation approaches worth mentioning.
Before we get to those, let's talk about the AI agent angle, because that's where Daniel's question gets genuinely interesting. He mentioned an agent that could handle the purchasing — you say "I need a wall anchor" and it does the rest.
This is the part where I get excited and then immediately disappointed. What would a truly autonomous shopping agent need to do? Step one, understand the product from natural language. "I need a keystone to RJ45" — that's actually a solvable problem. Large language models can parse that and identify the product category. Step two, find the right listing on AliExpress or Amazon. That means searching, filtering by price and rating, and picking a listing that's not a scam. Step three, add it to a shared cart. Step four, monitor the cart total against the threshold. Step five, auto-checkout with payment.
And each of those steps has a failure mode.
Multiple failure pattern. Let me walk through what's actually possible today. There's an open-source project called Browser Use that lets AI agents control a web browser. You give it a natural language instruction, and it navigates websites, clicks buttons, fills forms. There are demo videos of it adding items to shopping carts. It's impressive. But here's where it falls apart: checkout.
CAPTCHAs?
CAPTCHAs are the obvious blocker. AliExpress throws up CAPTCHAs during checkout, especially if it detects automated behavior. Browser Use agents can't reliably solve them. But even before CAPTCHAs, there's the problem of DOM changes. AliExpress updates its page structure frequently. An agent that worked yesterday might fail today because the "Add to Cart" button moved or changed its CSS class. Browser automation is inherently fragile against actively maintained websites.
And we're talking about real money here. The stakes are different from an agent that summarizes a webpage.
That's the crucial distinction. OpenAI demoed their Operator agent adding items to a cart, but they required human confirmation for payment. Nobody is letting an agent auto-charge a credit card without oversight, and for good reason. Imagine the agent misreads a listing and orders a thousand keystone jacks instead of one. Or it finds a listing that looks right but is actually a different connector standard. The error cost is real money, and the return process on AliExpress is not exactly frictionless.
So the fully autonomous AI shopping agent is a demo, not a product.
It's a demo that works in controlled conditions and fails in the wild. I'd say we're twelve to eighteen months away from agents that can reliably complete purchases on major e-commerce platforms, and even then, the trust question is bigger than the technical question. Would you let an agent spend your money without review?
I'm a sloth. I barely trust myself to spend my money.
But let me give you one more data point on why this is hard. Browser Use, the open-source project, has explicit documentation about handling dynamic websites. They recommend using visual element detection rather than CSS selectors, because visual detection is more robust against DOM changes. But visual detection is slower and more compute-intensive. And even with visual detection, checkout flows on AliExpress involve multiple redirects, payment gateway popups, and address verification steps that are different on mobile versus desktop. It's a hard automation problem.
So the AI agent approach is the destination, but we're not there yet. What about the middleware approach? A service that sits between the user and AliExpress, maintaining a persistent cart?
This is technically feasible. You'd build a browser extension that intercepts "Add to Cart" actions and stores them in a cloud database instead of, or in addition to, the AliExpress session. The extension would maintain its own cart state, synced across devices and users through the cloud backend. When the total hits the threshold, the extension would automate the checkout flow — filling in shipping details, selecting payment method, placing the order.
And this violates AliExpress's terms of service how exactly?
In every possible way. AliExpress's terms explicitly prohibit automated access. A browser extension that programmatically clicks through checkout is automated access. They can detect it, block it, and potentially ban accounts that use it. The extension would be playing cat and mouse with AliExpress's anti-bot systems indefinitely.
So it's a technically interesting idea that lives in a legal gray zone and would break constantly.
That's the middleware story in a nutshell. It's the kind of thing a motivated developer could build for personal use, but it's not a product you can ship to consumers. The maintenance burden alone would be enormous — every time AliExpress changes their checkout flow, you'd need to update the extension.
Which brings us to the practical question. What can Daniel actually do today?
Before we get to workarounds, I want to flag one more thing about the current state of AI shopping agents, because it connects to a broader trend. There's a wave of startups building what they call "conversational commerce" agents — you describe what you want in natural language, and the agent finds and orders it. But they all stop short of auto-checkout. They'll fill your cart and then send you a link to review and pay. The payment step is still human-gated everywhere.
Which is probably the right call for now. But it does mean that Daniel's vision of "add items as you think of them, order auto-places at threshold" isn't available as a product.
Not as a single product. But you can approximate it with a combination of tools that each handle one piece of the puzzle. Let me walk through three workarounds, from simplest to most technical.
Let's start with the one that requires the least setup.
Share-A-Cart plus a shared messaging channel. Daniel and Hannah install the Share-A-Cart browser extension. They create a WhatsApp or Telegram group called "AliExpress Cart." Whenever one of them adds items to their cart, they generate a Share-A-Cart link and paste it in the group. They designate one person — let's say Daniel — as the order consolidator. Once a day or every couple of days, Daniel checks the group, imports the shared carts, combines everything, and places the order if it's above ten dollars.
That's... actually pretty reasonable. The friction is generating and pasting the link, which takes maybe ten seconds.
It's the least elegant solution, but it works today. The key insight is that you're externalizing the shared cart into a messaging channel, and you're making the threshold check a deliberate human action rather than an automated trigger. It's not "set it and forget it," but it solves the cross-user problem and eliminates the mental overhead of remembering what the other person added.
And it doesn't violate any terms of service. You're just sharing links.
It's using the platforms as intended, just with a manual coordination layer on top. Now, the second workaround is for iOS users specifically.
Daniel and Hannah are both in the Apple ecosystem, so this one's relevant.
They create a shared Reminders list called "AliExpress Cart." When one of them thinks of an item they need, they add it to the list with the product name. Then they set up a Shortcuts automation that triggers when a new item is added to that list. The shortcut opens the AliExpress app with the search field pre-filled with the item name.
So it's not auto-adding to cart, but it's reducing the friction from "remember to search for this later" to "tap the notification and you're on the search results page."
Right. It's a nudge system rather than an automation system. The mental model shifts from "I need to remember to buy this" to "I've captured this, and the system will prompt me to buy it." That's a meaningful reduction in cognitive load, even if it's not the full automation Daniel wants.
And the third workaround is for the technically inclined.
This one uses n8n or Make.com — these are workflow automation platforms. Daniel sets up a shared Google Sheet where he and Hannah log items they want to order. Each row has the product name, the AliExpress URL, and the price. An n8n workflow runs daily, sums the prices, and if the total exceeds ten dollars, it sends an email or a Telegram message saying "Your cart is ready — combined total is twelve dollars and forty cents."
So it's automated threshold monitoring without automated checkout. The machine does the math, you do the clicking.
And that's actually the sweet spot for current technology. Automating the monitoring is easy and safe — it's just arithmetic on a spreadsheet. Automating the checkout is hard and risky. By splitting them, you get most of the cognitive benefit with none of the liability.
I want to pull on a thread here, because I think there's something bigger underneath Daniel's question. He's not just asking about AliExpress. He's describing a missing UX pattern across all of consumer e-commerce.
Say more.
The shared auto-threshold cart. It doesn't exist anywhere in consumer e-commerce. Amazon has Subscribe and Save, which is per-product auto-reordering. But that's for consumables you buy on a schedule — toothpaste every three months. It's not for the random assortment of small parts you need for projects. And it's per-user, not shared across a household.
And Amazon has the Amazon Household feature, which lets you share Prime benefits and payment methods across two adults and up to four children. But the carts are still separate. You can't pool items into a shared cart. It's baffling.
It's baffling until you remember what we said earlier. The friction is the feature. A shared household cart that auto-orders at threshold would reduce the number of browsing sessions. Each household member would spend less time on the platform. The algorithm would have fewer opportunities to surface recommendations.
There's an interesting counterpoint, though. AliExpress actually benefits from order consolidation. Daniel mentioned this in his prompt — consolidated packages are easier to pick up, and they reduce AliExpress's shipping costs because they're bundling multiple items into fewer shipments. There's a world where AliExpress could offer a shared household cart as a premium feature, because the shipping savings offset the lost browsing revenue.
But that would require them to acknowledge that cross-device cart sync is broken and build a proper solution. Which means investing in infrastructure that doesn't directly generate revenue.
The classic platform incentive problem. Fixing the cart sync doesn't increase sales in a measurable way, so it never gets prioritized. Meanwhile, the recommendation algorithm gets a team of fifty engineers.
And that's why the workaround ecosystem exists. Share-A-Cart, browser extensions, shortcut automations — they're all filling gaps that the platforms have no incentive to fill.
Let me add one more layer to this, because I think it connects to something Daniel's been thinking about with AI agents more broadly. The reason the auto-threshold cart is such a compelling idea is that it inverts the relationship between the user and the platform. Right now, the platform sets the terms — you browse when we want, you see what we show you, you check out when you're ready. An auto-ordering system says: I set the rules. You execute when my conditions are met.
It's procurement thinking applied to personal consumption. Daniel's essentially asking, why can't I treat AliExpress the way a company treats its suppliers?
And the answer is, because you're not a company with negotiated terms and API access. You're a retail customer, and retail customers are supposed to be passive participants in the platform's funnel.
That's the structural tension. But it does make me wonder about the long-term play. If AI agents become reliable enough to handle checkout, the platforms will have to respond. Either they'll embrace agent-friendly APIs, or they'll wage an escalating war against automated access.
I think we'll see both. Some platforms will offer official agent APIs with built-in spending limits — "you can auto-order up to fifty dollars per month without human approval." Those platforms will capture the automation-enthusiast segment. Others will fight it and lose users to the ones that embrace it.
The first platform that offers a "set your threshold, we'll auto-order" feature with sensible fraud protections wins a chunk of the market that currently cobbles together Share-A-Cart links and Google Sheets.
And the fraud protection piece is solvable. You could have a system where the auto-order triggers, but the payment is held for thirty minutes with a cancellation window. If you get a notification saying "Your AliExpress order for twelve dollars has been placed — tap to cancel," you've got a safety net. That addresses the liability concern while preserving the automation benefit.
So the technology for the safety net exists. It's the will to build the feature that's missing.
And that brings us back to the practical. Daniel's not going to wait for AliExpress to have a strategic awakening about shared household carts. He needs a system that works this week.
So let's summarize the actionable path. Share-A-Cart extension on both browsers. A dedicated WhatsApp group for cart links. Daniel as the consolidator. He checks the group every day or two, imports the shared carts, and places the order when the total crosses ten dollars.
And if he wants to reduce the mental tracking, the Google Sheet plus n8n workflow gives him a daily email with the running total. It's not elegant, but it's functional.
The inelegance is almost the point. The fact that you need three tools and a manual consolidation step to approximate what should be a built-in feature tells you everything about where consumer e-commerce is versus where it could be.
And it tells you why Daniel's instinct to ask about AI agents is the right long-term bet. The agents aren't ready today, but they're coming. When they can reliably navigate checkout flows, the shared auto-threshold cart becomes a simple prompt. "Monitor this shared list. When the total exceeds ten dollars, place the order using the saved payment method. Send me a confirmation."
The question is whether we'll trust them. And I think that's a bigger hurdle than the technical one. People are going to be deeply uncomfortable with software that spends their money without a human in the loop, even for small amounts.
I agree, but I think the comfort will come gradually. It'll start with low-stakes purchases — the ten-dollar AliExpress cart, not the five-hundred-dollar Amazon order. Once people see that the agent gets it right ninety-nine times out of a hundred, the trust threshold expands. It's the same pattern we saw with ride-sharing. Nobody wanted to get in a stranger's car, until they did.
Fair. And Daniel's use case is actually the perfect onboarding ramp for this. Small amounts, non-urgent items, clear rules. If an AI agent messes up a two-dollar keystone jack order, the downside is negligible.
The stakes are exactly right for experimentation. Which is why I suspect Daniel will be an early adopter of whatever agent-based shopping tool emerges first.
So to directly answer Daniel's question — does anything like the shared auto-threshold cart exist today? No, not as a single product. The enterprise procurement world has it, but consumer platforms don't. The closest workaround is Share-A-Cart plus a shared messaging channel, with optional spreadsheet automation for threshold monitoring. And the AI agent approach is demos and prototypes, not production-ready tools.
But the gap is closing. The fact that we can even have a serious conversation about browser-based AI agents completing purchases — that wasn't possible two years ago. The trajectory is clear, even if the destination isn't here yet.
And in the meantime, the friction you're feeling is a feature, not a bug. The platforms want you to browse. But you can hack your way to a better system with the tools that exist today.
The open question I keep coming back to is whether consumer platforms will ever embrace auto-ordering natively, or whether it'll always be a hack layered on top. AliExpress benefits from consolidation — fewer packages to ship, happier customers picking up one bag instead of five. There's alignment there. But the fraud and liability risk is a genuine blocker, and nobody wants to be the first platform to deal with the headline "AliExpress AI Agent Orders Ten Thousand Dollars of Keystone Jacks."
That would be a very specific news story.
Stranger things have happened. But I think the first platform that offers agent-friendly APIs with sensible auto-checkout limits and a cancellation window will capture a wave of users who are tired of the manual consolidation dance. Daniel's not alone in wanting this. He's just articulate about it.
And now: Hilbert's daily fun fact.
Hilbert: The only known surviving example of a Tasmanian mouth bow — a single-string instrument played by plucking with the mouth as a resonator, used by Aboriginal Tasmanians — was collected in 1903 and is held in the Tasmanian Museum and Art Gallery. The string was traditionally made from kangaroo sinew.
...right.
Thanks, Hilbert. That's a lot of kangaroo sinew to process.
Try the Share-A-Cart workaround this week. It's not the elegant solution Daniel wants, but it's the one that works today. And when the AI agents are ready to handle checkout reliably, you'll already have the workflow in place — you'll just be replacing the manual consolidation step with a prompt.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this episode, tell someone who's ever waited three weeks for a two-dollar part to hit a free shipping threshold. We'll be back soon.
See you tomorrow.
