What is the use case for LLM agent shoppers? I can't imagine delegating the purchase of a used item to an AI (I'd be okay with AI identifying the best deals for me to review). This must be something for people who are doing something at scale like flipping items on Ebay or drop shipping.
I imagine this type of automation existed before LLM agents came along - what do they add? Is it just the ability to evaluate the product description? Item quality is already listed as a categorical variable.
Hasn't eBay's traffic been 80% bots since day one? I haven't participated in an auction in forever because even 20 years ago you were guaranteed to get sniped by a bot on anything except actual garbage.
Tried selling on eBay as a regular Joe lately? Item sold for roughly $190 and I lost $45 in fees - I didn't even have a premium ad or pay for any of the boosting.
Interesting, I’m not big on AI but I have thought often it would be nice to have an ‘agent’ that monitors ebay or other classifieds sites for items based on a natural language description.
Something like “I want an old mini PC to use as a home server, it should have roughly these specs and cost under this amount”, and then an LLM would run some searches every day, parse the results and send me a message if something comes up.
It’s pretty easy to get alerts for when items are available for a certain price if you know the exact item you want, but on eBay and classifieds sites, I usually just want something in a rough ballpark, and the best way to find that is come back and check every day looking through searches.
I don’t really see any value in having the AI do the purchase itself though.
Does the tools and features ebay already has not meet this need?
Can't you set up a saved search that ebay will notify you of?
>“I want an old mini PC to use as a home server, it should have roughly these specs and cost under this amount”
This is a bad example because at pretty much all times, there is sufficient inventory for you to find the actual item you want, so you don't need the "agent" to repeatedly check. In instances where there is limited inventory, saved searches have been the reliable solution for decades. It's how niche youtube channels have acquired niche hardware forever.
I found this interesting to think about. On the surface, this could fit the requirements. However, in the past decade, I feel a dynamic where my relationship with tech companies like eBay are actually antagonistic, rather than cooperative. They (and American society) have become so extractive and user-hostile that I have no trust their features are designed to accomplish my goals. Instead, they are designed to accomplish the company's objectives and it's only by coincidence if the customer is satisfied.
>eBay and classifieds sites, I usually just want something in a rough ballpark, and the best way to find that is come back and check every day looking through searches.
You already can do that on ebay. Whatever terms get you ballpark results can become a search alert. You don't need to know the items you want exactly. And you get a handy email with the results of that alert when they hit, which you can scroll over in about ten seconds.
I thought botting on ebay has been forbidden since forever, LLM or otherwise. This isn't so much a policy change as an affirmation that existing policy still applies.
This ban isn't about 'fairness' or bot protection; it's about protecting the Impression Funnel.
Marketplaces like eBay are designed to monetize 'Wandering Attention'—sponsored listings, 'customers also bought', and sidebar ads.
An AI Agent represents 'Laser Focused Attention'. It executes a transaction with zero wandering. It effectively turns the marketplace into a commoditized backend database / dumb pipe.
From a Growth/Unit Economics perspective, an AI Agent is a nightmare customer. It has zero probability of impulse buying and generates zero ad revenue. They are banning them to save their business model, not their inventory.
Or probably even wrose, it actually shifts the attention and the wandering. That phase will happen inside the LLM, where the LLM decides which link to suggest, i.e. whoever pays the LLM the most. And worse yet, that will apply not just for products, but for platforms, so if amazon pays chatgpt more than ebay does, there goes your sale.
An AI-Agent browsing eBay for a "widget" for a given individual will also likely not be browsing eBay's advertising listings (sponsored and promoted listings [1]) which would potentially equate to a loss in ad revenue for eBay. So there is likely a "protect the advertising moat" aspect to their "ban" as well.
[1] Given how hard eBay pushes sellers to purchase the sponsored and promoted listing tiers (at an additional fee of course) implies they make some nice revenue stream from these advertisements.
I don't think the 'wandering attention' dissapears, rather it is pushed to the LLM product. It's more of a competitive transfer from the incumbent product category to the new one, it's not that the new product category 'fixes' the 'problem'
If I am able to find what I’m looking for and purchase it via AI why must I be subjected to advertising & promotion of items in have zero interest in? Amazon has made finding what I want painful. I suspect eBay is just as painful (I see to work there).
That's a cynical take, so it will probably get upvoted, but what are you basing it on?
Ebay is a pretty eclectic marketplace and I can think of a number of possible reasons that have little to do with ads. For example, they may be worried about high error rates, and thus buyer and seller dissatisfaction. If I instruct an agent to buy X, eBay is almost never interchangeable with Amazon or Target.
They have no problem surfacing their listings on Google Shopping.
But, ads directly correspond to revenue stream, and a loss of ad "impressions" would result in a reduced revenue stream, so a "protect the advertising" response is not at all unusual to consider as a portion of their (eBay's) reasoning for this ban.
Given how hard they push sellers to purchase their "extra cost listing enhancements" (i.e., purchase to have your listings show in the "advertisement" spots) it appears that they may make a decent revenue stream from these advertising angles. An AI-agent could find listings without going through the advertising displays and as such cut into this revenue stream.
Great analysis of the real motivation here. But this feels like the record labels trying to ban MP3 players. You can protect the impression funnel today, but the trajectory is clear - consumers will increasingly delegate purchasing decisions to agents, and the platforms that adapt will capture that flow.
The marketplace that builds "agent-friendly" commerce (verified listings, structured data, transparent pricing, API access) becomes the default backend for AI shopping. The one that bans agents becomes a legacy system humans have to manually navigate when the agent can't help.
eBay's current business model may be a "nightmare customer" for AI agents, but that's a problem with the business model, not with the agents.
This is a case where it may be that people are outsourcing shitty user experiences to an AI.
I’m not a huge ebayer but I’m usually watching one or two auctions at a time. The problem is that you can’t disallow marketing notifications. So, if I want to be usefully alerted for a new item in a search, or that I’ve been outbid, or the imminent end of an auction, I’ll also be getting notifications and emails about all kinds of shit I don’t care about. $5 off coupons (that only apply to 8 items that I don’t want). “You might like this!” notifications (spoiler: I never do). Group buying times (who cares?).
So I either disable ALL notifications (and have an LLM write a script that crawls searches manually and much more appropriately notifies me on its own), or I enable notifications and get a bunch of trash spam.
As it relates to specifically to buying, we’ve known for a long time that we’re all up against some kind of bot that’s timed the exact last moment and amount to outbid us. It’s no fun.
I’ve been an eBay user since 1998 and it’s been on a very slow roll of enshittification since then.
Make your experience better for humans and maybe we’d be less inclined to outsource negative experiences to AI.
Can’t charge for something if you’re giving it away for free.
Data’s the only moat left. Companies like stack overflow need to build revenue streams from AI or they will cease to exist.
By banning bots and then licensing some kind of access, eBay can protect itself from merely being a listing point that no human actually visits. Tailwind and their adverts via docs model, eBay and its promoted listings model, we’re going to see businesses adapt or die on this.
It may just be to stop third parties from creating a whole business out of "shop for me" AI bots. Individual users getting away with it might not be a problem, but with it being against ToS, it'd be a lot more shaky to build a business out of it.
In fact, it may just be that eBay wants to be the business selling AI "buy for me" agents.
example of focused leadership - a commenter already noted how wondering listings drive revenue
if it was some Bozo executive as we see at most tech companies - they would be advocating to implement the Open Agentic Commerce whatever being pushed by google while not noticing its killing their own company
Dumbest thing in the world, not wanting buyers who are ready to complete transaction the moment they find what they want. When my car broke down and cost too much to repair I described what I need - low mileage, big trunk, 40+ MPG, under certain price, close to dealership where my broken car sat in service. That I gave the query to Grok 4 Expert (because it does heavy web scraping), found a 2020 Ford Escape Hybrid and drove it away two hours later, because rationally speaking missing a lot of work in this economy is a bigger financial risk than missing $3000 on manual bargain hunting vs a good AI hunt. If any dealers blocked scaping on sales pages, too bad for them. Speaking of which, any good bargain / secondhand market AI agent friendly API?
And we enter the predicted cycle of "new thing that's going to Connect Everything"
Just like faxes, the internet and the world wide web, sure, this new thing Could connect everything, but that's not nearly as much of a technical problem as a sociopolitical one. Same as it ever was.
58 comments
[ 2.4 ms ] story [ 80.8 ms ] threadImpossible to enforce, they can read browser windows and pass captchas
I imagine this type of automation existed before LLM agents came along - what do they add? Is it just the ability to evaluate the product description? Item quality is already listed as a categorical variable.
No wonder Facebook marketplace has destroyed them
Something like “I want an old mini PC to use as a home server, it should have roughly these specs and cost under this amount”, and then an LLM would run some searches every day, parse the results and send me a message if something comes up.
It’s pretty easy to get alerts for when items are available for a certain price if you know the exact item you want, but on eBay and classifieds sites, I usually just want something in a rough ballpark, and the best way to find that is come back and check every day looking through searches.
I don’t really see any value in having the AI do the purchase itself though.
Can't you set up a saved search that ebay will notify you of?
>“I want an old mini PC to use as a home server, it should have roughly these specs and cost under this amount”
This is a bad example because at pretty much all times, there is sufficient inventory for you to find the actual item you want, so you don't need the "agent" to repeatedly check. In instances where there is limited inventory, saved searches have been the reliable solution for decades. It's how niche youtube channels have acquired niche hardware forever.
You already can do that on ebay. Whatever terms get you ballpark results can become a search alert. You don't need to know the items you want exactly. And you get a handy email with the results of that alert when they hit, which you can scroll over in about ten seconds.
Marketplaces like eBay are designed to monetize 'Wandering Attention'—sponsored listings, 'customers also bought', and sidebar ads.
An AI Agent represents 'Laser Focused Attention'. It executes a transaction with zero wandering. It effectively turns the marketplace into a commoditized backend database / dumb pipe.
From a Growth/Unit Economics perspective, an AI Agent is a nightmare customer. It has zero probability of impulse buying and generates zero ad revenue. They are banning them to save their business model, not their inventory.
[1] Given how hard eBay pushes sellers to purchase the sponsored and promoted listing tiers (at an additional fee of course) implies they make some nice revenue stream from these advertisements.
Ebay is a pretty eclectic marketplace and I can think of a number of possible reasons that have little to do with ads. For example, they may be worried about high error rates, and thus buyer and seller dissatisfaction. If I instruct an agent to buy X, eBay is almost never interchangeable with Amazon or Target.
They have no problem surfacing their listings on Google Shopping.
Given how hard they push sellers to purchase their "extra cost listing enhancements" (i.e., purchase to have your listings show in the "advertisement" spots) it appears that they may make a decent revenue stream from these advertising angles. An AI-agent could find listings without going through the advertising displays and as such cut into this revenue stream.
The marketplace that builds "agent-friendly" commerce (verified listings, structured data, transparent pricing, API access) becomes the default backend for AI shopping. The one that bans agents becomes a legacy system humans have to manually navigate when the agent can't help.
eBay's current business model may be a "nightmare customer" for AI agents, but that's a problem with the business model, not with the agents.
(Generated comments aren't allowed here - we only want commenters who write in their own voice. More explanation at https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que... and https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...)
I’m not a huge ebayer but I’m usually watching one or two auctions at a time. The problem is that you can’t disallow marketing notifications. So, if I want to be usefully alerted for a new item in a search, or that I’ve been outbid, or the imminent end of an auction, I’ll also be getting notifications and emails about all kinds of shit I don’t care about. $5 off coupons (that only apply to 8 items that I don’t want). “You might like this!” notifications (spoiler: I never do). Group buying times (who cares?).
So I either disable ALL notifications (and have an LLM write a script that crawls searches manually and much more appropriately notifies me on its own), or I enable notifications and get a bunch of trash spam.
As it relates to specifically to buying, we’ve known for a long time that we’re all up against some kind of bot that’s timed the exact last moment and amount to outbid us. It’s no fun.
I’ve been an eBay user since 1998 and it’s been on a very slow roll of enshittification since then.
Make your experience better for humans and maybe we’d be less inclined to outsource negative experiences to AI.
Data’s the only moat left. Companies like stack overflow need to build revenue streams from AI or they will cease to exist.
By banning bots and then licensing some kind of access, eBay can protect itself from merely being a listing point that no human actually visits. Tailwind and their adverts via docs model, eBay and its promoted listings model, we’re going to see businesses adapt or die on this.
In fact, it may just be that eBay wants to be the business selling AI "buy for me" agents.
if it was some Bozo executive as we see at most tech companies - they would be advocating to implement the Open Agentic Commerce whatever being pushed by google while not noticing its killing their own company
Just like faxes, the internet and the world wide web, sure, this new thing Could connect everything, but that's not nearly as much of a technical problem as a sociopolitical one. Same as it ever was.