Launch HN: Channel3 (YC S25) – A database of every product on the internet
Hi HN — we’re George and Alex, building Channel3 (https://trychannel3.com/), a database of every product on the internet, searchable via text/image, and with built-in affiliate monetization. Here’s a demo: https://www.youtube.com/watch?v=Mx8FyP7KvJg.
It’s surprisingly hard to find good product data. If you want your software to recommend products and deep-link to merchants, you’ll quickly discover that the data you need—clean titles, normalized attributes, deduped listings, current prices and inventory, variant options, images, and brand info—is not just messy; it’s also spread across a long, long tail of retailers, and often lives behind advanced bot-detection systems.
We ran into this problem while building an AI teacher that could recommend relevant supplies. We asked Exa for products, but got back articles, not structured data. Same for Tavily and Bing (deprecated as of 8/13/25). And we got rejected from affiliate programs, who suggested we come back with 1000s of TikTok followers. Channel3 is the API we wished we had.
Product detail pages (PDPs) usually present the main item alongside recommendations. We use computer vision to isolate the main product and find its attributes, like title and price. We apply the same logic to the rest of the PDPs on the domain.
Products are often sold across multiple retailers, with no guarantee they’ll be labeled consistently. We collapse products across the web into a canonicalized set by using LLMs and multimodal embeddings to actually understand each product.
To normalize everything into a schema that tries to be both minimal and extensible, we have to be opinionated. Are a $50 10” and $60 12” skillet the same product? Probably not, but the S/M/L variants of a T-shirt are. Our goal is that any product you’d search for specifically is treated as its own product.
We process a massive amount of data. We quickly ran out of room on our Cloudflare Vectorize indices and moved to the brand-new AWS S3 Vectors platform, syncing to OpenSearch for faster response times and more dynamic filtering. We hit rate limits constantly, so we spread our work over multiple cloud providers and AI models.
You can search things like “outdoor grill, less than $1000”, or “sweat-resistant, wireless running earbuds”, or "women's jeans from Paige that look like [https://www.gap.com/webcontent/0020/666/799/cn20666799.jpg]”. Products come back as JSON with title, brand, images, price, specs, etc.
Developers earn commission on sales they drive (averaging 5%). Channel3 takes a cut. We want you to earn way more money from Channel3 than you spend on it. We win when you win.
We provide an API, SDK (Typescript and Python), and MCP. We offer 1000 free searches, and charge $7/1000 searches after that. You can view expected commissions per-brand on our dashboard.
So far, products are US-only (sorry! we will expand). We’re live with millions of products and hundreds of developers.
To get started, make a free account at https://trychannel3.com, then select which brands you’d like to sell (choose from 50k+ or request your own), generate an API key, and start selling and earning.
We’d really appreciate feedback from this community. If you’ve built product search before, what did we miss in the schema? If you’ve tried to add commerce to an app, what blocked you? If you tried to build this yourself, what did you learn? Are there endpoints you wish existed (e.g. price alerts, back-in-stock webhooks, product feed)? For any suggestions, we’re all ears.
We’ll be in the thread all day to answer questions, share more technical detail, and hear whatever would make this most useful to you. Comment away!
I assume that payments from purchases come from you guys, rather than me needing to create and manage an affiliate account with each individual vendor?
You say that commissions average 5%, but what is the variability and where does it come from?
Last, a bit of feedback about the product.
I tried searching "nintendo switch 2" on your homepage and the results that came up kind of sketched me out. You mention that the products are US-only, but the first result clearly says "hong kong" in the title. And the store listed is "My Nintendo Store PT"; is that the official store? When I google that it takes me to the Portuguese version of the nintendo website, and that makes me even more confused.
The second result for the same search appears to be a dress, which is obviously completely unrelated to video games in general.
EDIT: I'm noticing irrelevant results for many queries. Searching "plain white pillowcase", the third result is a t-shirt, the seventh result is a dress, and the eleventh result is a light bulb.
Searching "men's wallet" the very first result is an outdoor picnic table.
Regarding payments, your understanding is correct. We have and manage our affiliate partnerships, all you have to do is drive sales and we forward on the commission to you. We're working on improving signal into the range of commissions you can expect, but, in short, the variability stems from merchants and product type. For example, technology (e.g. iPhones, laptops) typically have lower commissions than beauty supplies.
Thanks for the feedback. Managing and cleaning this volume of data is an ongoing task, and our catalog is getting better each day. I'll check out the nintendo case in particular.
Putting my affiliate hat on here for a minute...
Very cool to see how you've aggregated so many products into one service. How do you plan to compete with FMTC and others that aggregate feeds together? Speaking as a publisher, I'd not want to share commission unless absolutely necessary and would prefer to just pay a fee so I can access the feed and not have an unknown amount of revenue lost between myself and the merchant.
As a brand running a program, I'd be very cautious about allowing my feed into your database if I didn't have any way to finding out who is featuring my products and where/how. Are you providing visibility to the brands since you're effectively functioning as a sub-affiliate network?
Those questions aside, great to see YC funding a startup in the space!
Yes certainly! I've dealt with large datasets like this in the past and know firsthand how challenging it can be to wrangle them.
Something like this would be a great fit for my travel planner app if I knew I could trust that the results were high quality before prompting the user with them.
Btw I edited my earlier comment with a few more examples just before you replied.
Good luck!
Appreciate it. FYI, for the specific bugs you flagged, looks like Nintendo was improperly named (reindexing products with that name now), and sounds like the pain point you felt was extraneous search results that really didn't belong. Transparently, the problem we're facing there is vector search can be a bit of a black box, so we're trying to tune our hybrid search to cull out really crazy results, but obviously it still needs work.
One of the ways we're combatting these search problems in the early days is developers can curate their catalog with specific brands, merchants, and categories (and even down to the product level) so you know exactly what the search space for each of your queries is. Curious to hear about your travel planner app -- if you think this would be a helpful tool, feel free to reach out at george@trychannel3.com
Hi, we fixed this bug and you should see better results now. I'm getting good results for "men's wallet" and "plain white pillowcase". Try it out and let me know what you think!
The results look much much better now. Great job.
I am confused on who this is exactly for ? Is it for an end user/buyer who is looking to buy something and you just aggregate from various sources based on query OR is this for other developers/providers who are providing their own search interface on their own eCommerce website etc ? I assume the latter but isn't very clear at least to me.
I was confused at first as well, but I think I kind of get it.
I have a site that has lots of sports data on it. If you go to look at a page for one of the teams, maybe I want to put a link to buy their jersey. I could potentially go direct to the source and apply for affiliate programs from Nike, Adidas, Under Armor, Puma, etc, but that's a lot of effort to both set up and maintain across all the different partners for something that's not part of my core business.
Instead I can just use Channel3's affiliate links. I make a little less, but the trade off might be worth it for the simplicity.
Honestly, I'm not 100% sold on the idea being a good business, but I kind of get the idea. My concern would be maintaining the whale type customers. It feels to me like the product makes sense for minnows who don't have high volume or for people building an MVP or looking for a proof of concept, but once a customer has hit a certain scale, I don't get why you wouldn't just get a direct affiliate program and cut these guys out.
If I'm driving a $1000 a month to Nike and getting 3% from these guys, but I can get 5% by setting up an affiliate program directly with Nike, maybe it's not worth my time to chase that extra $20. If I'm driving $50k a month to Nike then that's an extra $1000 a month I could get by cutting these guys out which feels a lot more worth it.
Developers! Anyone who wants to add shopping to their platform, build an e-comm website, or monetize their agent can use Channel3 to earn commissions on the products they sell. Totally see how that could get lost in this post, we tried to focus more on what we built than try to sell to devs. Hopefully our website makes this clearer.
Can you be more specific?
Can you explain the main use cases when i would want to add shopping to my platform?
If i’m building an ecommerce website, why would I need your API if I’m just selling my own products?
Yep, I should be more precise
> If i’m building an ecommerce website, why would I need your API if I’m just selling my own products?
You're right, Channel3 isn't for existing ecommerce websites. Channel3 let's anyone build a shopping experience (which I vaguely conflated with ecommerce website). You don't worry about managing product, you just build the platform. Some AI shopping experiences like this already exist, check out plush.shop, daydream.ing, and onton.com.
> Can you explain the main use cases when i would want to add shopping to my platform?
In short, if you want to monetize your platform without running ads. What's neat is there aren't really "main" cases -- this is up to the inventiveness of our users! We believe some of the most lucrative opportunities are yet to be imagined. My co-founder, Alex, experienced this problem at his last job when he was building an AI tutor; they decided to try to add an additional revenue stream by letting the AI tutor recommend products. Maybe blogs can integrate an AI-recommended product feed based on their article. Maybe yoga teachers who have a website for booking classes can recommend their gear to their students and earn some money when they do. Maybe someone just loves the color orange and wants to build a shop for orange products. We don't know what devs have in store for Channel3, but we do know agentic commerce is going to reshape how we interact with products!
So your target audience to be more precise are developers building ecommerce search engines, publishers and content creators.
I still fail to see a positive use case for this. Feels like offering flooding the ocean with toxic slop as a service on the presumption a few of the more lucky customers can pick up some tiny fish corpses floating to the surface.
[dead]
"Every product on the Internet" - "US-only, sorry!" ... Guess it's actually not every product on the Internet, not even remotely. Is it even 1 % of all products on the Internet?
Curious, how are you keeping the product data up-to-date? We built something similar for price alerts on specific URLs, that we use all the time, but have to poll it daily to see the price change (https://lowlow.bot). I imagine that would be a lot of $$ for every product on the Internet.
This is definitely one of our hard problems. There are some optimizations -- e-tags / last modified headers, comparing page content hashes -- but there's also only so much you can do before you just have to check the page again.
Last modified headers is always set to the current time in the majority of cases, and it also requires a web request too (albeit a HEAD request would likely suffice)
Seems like a great idea but the search is horrible. I tried a few ones and products were completely unrelated, or if related (because of a keyword) completely different from what I wanted. I understand you are working on the dataset, etc., but releasing too early may be bad for you guys. Most people won't try it again if the quality is bad. You have one chance to make an impression.
Thanks for the feedback. Releasing early vs releasing a polished product is always a tough balance to strike, and I'm grateful for insights like this. Do you have specific queries that missed the mark, or specific problems with the results? The common refrain we've heard is some products surfaced are good, but some random products also slip in there. If that's not consistent with your experience, I'd love to know what went wrong!
Hey! I know that you probably don't have any groceries there, but i tried "Goat cheese" and search result is super weird.
We build something similar and experience the same problem. Our current solution is not that effective, but we use vector-based search + llm to resolve such queries
Yeah, physical products for now. No groceries! We also use vector search + lots of LLMs, but not surprised you'd see funny search results here. We're working around the clock to make this better.
Hi, we just pushed up a fix. Looks like our vector results were acting up. Please let us know if you're getting better results!
Tried searching for "red high top sneakers under $40", and the first result was a shaft seal? The next result was a red high top sneaker so good, but it had a price of '0', bad. Then it started showing basically random items skin care, dog food, white sneakers, dog bandanas etc. The price limit started to get ignored.
Thanks for the feedback! We're working to improve search results right now. Bummer the price filter got ignored, we rolled out natural language filter extraction recently, clearly we have a bug there :(
Hi, we just pushed up a fix to search. You should see much better results for "red high top sneakers under $40".
The price errors stem from a bad sync between our product database and OpenSearch index. We're working on fixing this now. Let us know what you think!
I would presume you replied to the wrong user
Also, I'm always mystified in these Launch threads how much very simple QA gets pushed down onto users, or early adopters as is the case in these threads. How would you have spotted that bad sync if you didn't have someone giving you free QA in an Internet forum?
Oops! Responded to the wrong comment.
It's just two of us and we're working 120 hour weeks... don't have time for everything! We love feedback, it's super helpful.
I would actually say that the selling point here is that affiliate revenue goes thru your website.
I had tried to sign up for affiliate sales a while back, but:
It is complicated to sign up for it – depending on the vendor you have to fill in a number of forms, or sign up via a different affiliate network to even use them.
Wait times for a response are long – I remember some networks or individual sellers got back to me months later.
There's a high bar to entry – I had a tiny website, so I didn't get approved, but I had a good CTR. I eventually had to shut down the website since I realized there was no viable option to monetization and was just burning money on name registration + hosting.
My website was also not in the blog-space, i.e. I didn't do reviews, but I did offer good info, and Amazon for example specifically denied me affiliate permissions because of this.
I might revive the website and see if it'll work again with you guys. This is a path to monetization that could make it sustainable. Thanks and good luck!
Love this! This is one of the key painpoints we're trying to solve. It takes building in the space to know how hard (/impossible) monetizing products on your site is, but we think once people try to do so, they'll end up building with Channel3!
Excited to hear about what you were building. If there's anything Channel3 can do to support, feel free to dm me at george@trychannel3.com.
Congrats with the launch and good luck! I'm sensing (not speculating on when) a future where AI companions are integrated with all products in reach, so why not?
I'd be weary of paying too much attention to cynicism on here. However, as another guy mentioned, also great feedback and room for improvement mentioned. Cheers
Thanks for the note!
One of our clients is a procurement marketplace. One of the current struggles we face is getting vendors to upload catalog details for each product, which our marketplace needs to populate for a better shopping experience. Would your API assist us with filling in those gaps on products? (think common office products, industrial equipment etc). The caveat is that we can only show products from specific compliant vendors.
If a quick POC is useful, we (everfind.ai) can help fill product-data gaps while honoring your "compliant vendors only" rule. We ingest whatever you have (text, JSON, HTML, Excel), map it into category-specific schemas, and surface clean fields for search/filter. You also get fast fielded search, sensible relevance, hybrid vector search, and an optional guided-selling assistant you can expose publicly. Happy to try this on a small sample.
Feel free to reach out to me at felix.faust@everfind.ai :)
Certainly! Feel free to reach out to me directly at george@trychannel3.com with specific vendors.
How do you plan to differentiate vs. prompting a foundation model provider (interactively or via API for an affiliate site) with, "show me outdoor grills, between 500 and 1000, from weber"? https://imgur.com/a/Vdw4E1S
Certainly, for end users, foundation model providers (just ChatGPT right now) will be a great option for shopping. But, we don't believe it'll be the only place people shop. With Channel3, developers can build their own agentic shopping experience, and they also can monetize it.
By "via API for an affiliate site", I was referring to developer uses cases where the result can be used in a traditional website, by an AI agent, as a "related products" mechanism in an already existing system, etc. in order to monetize it. In that case, what will differentiate you?
Ah, sorry for misunderstanding. There are a bunch of challenges with that approach that my co-founder experienced first-hand (actually, that exact workflow is why we started Channel3!). To get up-to-date info, you'd need realtime websearch -- that's slow and expensive. To monetize, you'd need to set up affiliate relationships yourself. And, at the end of the day, the info you get from foundation models isn't really sufficient for building a rich shopping experience. So, someone could try that approach, or they could just use our API for cheaper and not deal with the hassle.
Great, thank you and good luck!
I'm curious if it's possible to take shipping prices into account. A $20 product with $30 shipping is a much worse recommendation than a $30 product with free shipping.
Exactly! We hope to be able to do this soon.
Are Amazon products available in your database?
Not yet, we're working on it though
So by definition not every product on the internet. Do you have my girlfriends handmade bags on there? I don't see them.
We're adding millions of products every week, so hopefully if you don't see your girlfriend's handbag there today, you'll see it soon!
i don't see the scarves that my cat sylvia knits and sells on etsy, either
no way. how are you not gonna get sued by them?
Searching for Ryzen 9950X3D yields 6 different AMD processors before I see a fully built PC with that Ryzen 9950X3D included. The actual product I was looking for was ranked 16th.
Thanks for the feedback! We're working on improving search, and it's helpful to hear the different ways it's currently coming up short.
$7 per 1000 searches seems steep. Don't think I will sign up to something where I have no idea if I'll make money, considering I'll also be sharing the commission with you.
Thanks for the feedback. We felt that $7 is reasonable. With an average online order value of $180 and average commission of 5%, you need to make just 1 sale in every 1000 to earn.
yes, really expensive. I would pay something like USD 0.01 / 1K requests... and even then, privacy and data ownership (even of queries themselves) is serious concern. especially you are just scraping other websites and re-selling their data. so why would I trust you with my data and funnel then?
Would you share more about what data you'd be concerned about here?
all of it
Wow. That steep indeed
> and often lives behind advanced bot-detection systems.
for a purpose. they own that data. now you are scraping and re-selling it. how you think they will like that? expect court orders and cease and desist letters coming in.
We've spoken with many merchants who are super excited about this use-case, actually! We think the web is opening up – e.g. Exa, Parallel, OpenAI, Anthropic, Perplexity, etc. all do the same, and it seems like this agentic future demands it.
OpenAI and Perplexity is notoriously getting sued by NYT, newspapers in Japan.
all do the same, which may be the reason why web is going to "close up", to prevent leaking data like this.
might be selective bias. try talking to folks who blocked scraping and you scraped them anyways and resold their data. see how happy they are. or try to talk to amazon, ebay, facebook for scraping their data, see how happy they are too.
yikes.
I can't see this disrupting the EDI world any time soon. Big business people will know what I mean.
All good, not the space we're targeting
I've been working with Alex at Channel3 closely on building a product on top of their service. They are very responsive and the product is getting better every day. I'm looking forward to seeing where Channel3 goes from here
We love to support our developers! Excited to see your progress
https://www.gs1.org/services/gdsn/global-data-model
As someone who has worked in e-retail, this catalog seems to have a lot of momentum.
Thanks for sharing. Always a fan of standardization of information online.
Nice! Reminds me of an old school YC co called Semantics3. Any connection?
haha, no, hadn't heard of them until now. uncanny resemblance though!
That's a crazy coincidence! Had no idea.
2nd employee at Semantics3 here. Considering all the AI available today I think things like product disambiguation becomes wayyy easier. We were trying many tricks and heuristics to identify the same products across sites.
Hey Shawn!
Sup!
Hi - we should chat :)
Man this is awesome. I will try it for Savely! (a price comparison browser extension)
Awesome!
Great idea but the search is awful which is dissapointing given how mainstream search has gotten.
Sorry to hear that! We're working to make search better, but we still have a long way to go. I'm curious where you felt the biggest pain points with search.
I think there is something wrong with your face filter settings on all your videos.
Too handsome?
Good idea. Business model seems pretty sketchy.
Rye allows you to even enter shipping info and pay for the product. But I can't help but feel it hasn't got the traction it deserves.
1) Are there plans to allow Devs to do the same?
2) why wouldn't you open the limit beyond 1000 free as long as you are making a rev share
3) does this pick from Shopify products/stores?
Imo the agentic loop isn't really closed unless you allow agents to pay and paywalls today aren't agent friendly. Tokenized cards, 16 digital cards. Perhaps but this involves high trust from users. Which means you are left with guiding users to the link and hoping they buy the product.
4) partnering with merchants where cards are already tokenized maybe your best converting potential customer base. But it's easy to do evil here, or loose trust without guardrails.
5) I would come up with a process to incentivize adding products to the ecosystem. However tiny the reward.
Lot of opportunity. Nice pitch. Good luck!
Would try this out of you can increase the 1k limit to something that is a win:win
Rye is an awesome company! They're in the universal checkout space now, stay tuned for a Rye + Channel3 demo in the next week :) We think the combo of product discovery with Channel3 and universal checkout is the future of commerce. To answer your questions:
1. Channel3 doesn't support shipping/payments, but there are a lot of great companies that do, so Channel3 + universal checkout is a full-fledged e-comm site. 2. We hope our pricing ($7/1000 req) is low enough that any reasonably-converting store won't need to worry about it. Average e-commerce order value is $180, so at 5% commissions you need <1 sale per 1000 queries. We're cheaper than any alternative, and, with rev share, we pay you to use us! (+ vector store is expensive :) ) 3. We do have shopify stores on our platform 4. An interesting idea! We're excited to see how agentic commerce evolves, and for now we're just trying to build the best discovery solution out there.
so you are scraping other vendors and online marketplaces? aren't you afraid getting sued by them? (e.g. by amazon?)
physical products, I couldn't find any software or SaaS
Are you using a browser provider or building your own CUA scraping pipelines in-house? What's your strategy around CAPTCHAs, paywalls, rate-limits, etc.?
All in-house! We don't visit sites with paywalls, but we do take lots of measures to avoid rate-limits.
Not seeing product country of origin info?
We don't currently surface that info
Hey George and Alex. This looks awesome. We're working on something similar, but for all of the businesses in the world: https://savvyiq.ai. We're international and have 265M+ entities in the system. We're actually preparing to do our own formal share on HN shortly.
We're working with enterprise customers now that want to use our system to dedupe all their gnarly business data, ground it to real legal entities, enrich it with base insights, then are asking for further data points more from a risk and due diligence standpoint.
Product information has come up repeatedly, but as you clearly know, that is a beast in itself that I don't think we'll ever tackle. For context, I helped build out the product data infra at https://www.wiser.com, and I'm not inclined to spend my time categorizing and building the taxonomy for pots, pans, and towels again.
I'm going to try out the product and happy to chat further if you think there's an opp to collaborate in some way. My email is in my profile.
Thanks for the comment. Will shoot you an email!
Prosperent returns
this is a cool idea. tried a few searches that were not confidence inspiring: missing prices, blurry, weird photos.
the searches in case you want to take a look: “dog collar” and “cat scratching post”
good luck with the product!
Thank you for the feedback. I tried "dog collar" and see what you mean -- looks like the blurry photos are super-zoomed-in headers from the website.
For this product (trychannel3.com/products/6PYSvW8), looks like we have a good photo for it, but also grabbed a couple we shouldn't have.
We're working around the clock to make this better and will add an AI-check to make sure the images we select are correct.
Hope you can check back soon and see some progress!
If I were to sell 10 items from 10 different brands, is the checkout process unified when the user as multiple items in the cart? As in, does the user have to checkout 10 different times at 10 different vendors?
Yes, but something like Stripe Order Intents (https://www.youtube.com/watch?v=bVQwIZYk9UM) would obfuscate this poor UX and make it really easy!
Would that interfere with trychannel's payment handling? I guess my use case isn't really what this is for.
dangerous business. when it does not work, nobody cares. when it works, you gonna get sued by markeplaces/vendors you were scraping (e.g. amazon, etc.).
[dead]