Did you validate this by running a A/B test? Main question is were you able to classify back into your known categories correctly all the time, or did the errors compound from the llm hallucination plus embedding search
softwaredoug•Aug 14, 2026
Using a Nano model, a tad worse than shipping a vocabulary to a larger OpenAI model. (And it’s an huge improvement on not classifying the queries at all).
But no classification is perfect. In search in particular, you will also want to have places for manual intervention for high priority queries.
VladVladikoff•Aug 14, 2026
Eh, maybe you should keep both paths. When LLMs eventually crawl the site to feed back to agentic shoppers, maybe they logically follow the more truncated less decorated path.
estetlinus•Aug 14, 2026
I was in a project where we sent the whole taxonomy every request, 40k tokens + one article, ”plz classify”. This was before structured outputs. It was extremely expensive and still hallucinated. Good ol’ days.
thatjoeoverthr•Aug 14, 2026
Smart! I've done the same trick for resolving extracted intents to selection.
But if accuracy matters, you can't rely on embedding sort to get a closet match. With a real test set they usually don't hold up under scrutiny.
Everything in AI is like this. You get an idea, try it once or twice, "LGTM" and you ship. Then it never survives contact reality.
Embedding sort gives you a better shortlist than the whole list, but you will probably want a heavier model to vet candidates.
piterrro•Aug 14, 2026
I would propose the following, query vector store for 10 closest categories based on a query, feed it to an LLM, in the prompt ask it to produce a single digit 0-9 representing the number of the most appropriate choice. Use plain text prompt, dont inflate token count with JSON.
There you go, you just drastically reduced the output pricing.
Additionally you could experiment with a reranker instead of an LLM or after reranking take top-3 results and then feed to LLM as input in order to reduce input token costs.
cesargstn•Aug 14, 2026
good this yeah
jddj•Aug 14, 2026
Or press 9 to hear these options again
virgil_disgr4ce•Aug 14, 2026
Your call is important to us. Please listen carefully, as our menu options have changed.
Colegno•Aug 14, 2026
Isn't search engines quicker than calling a LLM ? It might have a huge impact between a 20ms search engine call and a 2s LLM call for the end user.
fastball•Aug 14, 2026
A 2s LLM call is pretty slow.
gadflyinyoureye•Aug 14, 2026
Try using Digital Ocean. Minutes spent on inference.
quixoticaxolotl•Aug 14, 2026
They are already solving the problem with search engines, they're just using an LLM as a first pass to create better embeddings to run a similarity match on first. The difference in latency is likely made up for in accuracy.
amelius•Aug 14, 2026
Can anyone explain why LLMs are so bad at finding products (their webpages) with given specifications?
You'd think they would have solved it by now.
braiamp•Aug 14, 2026
Because that's structured data and structured data is usually hidden away from users _and_ machines. Product rarely want to be honest, unless it's B2B in a very competitive market (and even then!). So, yeah, it's not that they are bad, it's that there are few good sources of information.
(Lets ignore for now that no one seems to agree to what should be the spec sheets)
amelius•Aug 14, 2026
An LLM can read websites, right? And turn them into structured data.
ashu1461•Aug 14, 2026
It can do that on run time, but it does not store data like that. The data is typically stored as embeddings in which it is hard to query data in a structured form. Example give me all products whose price is less than 200$ vs suggest me products for my spouse's birthday.
amelius•Aug 14, 2026
Then they shouldn't store the data as embeddings.
Instead: use an LLM to build a large (old-school) database of products with all their specifications. The LLM can also build the schema for that database as it finds more data.
Then use an LLM to query that database based on the user's specifications (+ add some intelligence to find nice suggestions for a birthday if wanted, but I'd consider that an extra).
ashu1461•Aug 14, 2026
With agentic commerce protocol / unified commerce protocol open ai and gemini are trying to solve this problem.
The idea is to make structured queries using these protocols which can be used to fetch top products matching the user needs instead of just relying on semantic search.
It gets worse: shopping agents are hostile adversaries to Amazon unless they're paying Amazon and they've agreed to be friendly agents. No agent that won't betray you to an Amazon pricing strategy is going to be allowed access to Amazon structured data. They might even be fed poisoned data to discredit them.
But you'll be amazed by the abundance.
wslh•Aug 14, 2026
Because the data, in general, is not included in the LLM model and it needs to search/browse for external information. It cannot look indefinitely so it get the top results from lists, not "evrything".
_flux•Aug 14, 2026
Amazon Rufus has been mildly successful for me. I think the failures I've experienced with it are mostly because the product I'm looking for doesn't exist in the catalog.
pydry•Aug 14, 2026
It's been an absolute fucking disaster for me. It hallucinates endlessly and its searches are terrible. It even managed to confidently gaslight me about there being a VAT invoice available for a specific product.
I noticed yesterday when browsing on mobile that there used to be a box where I could search reviews and it got swapped with a Rufus box. I guess somebody needs to juice their engagement numbers for an investor briefing.
honestly, Amazon doesnt even need AI it just needs a better UI, more metadata for its products and to make reviews less scammy.
sgc•Aug 14, 2026
I asked a question once and now there is a effing alexa for shopping toolbar that takes a quarter of the screen that will not go away no matter how many times I close it, and the space remains taken even if I adblock it. Absolutely hostile implementation. I have words for this I cannot type out.
simonw•Aug 14, 2026
LLMs aren't architected to handle filter-style comprehensive search without setting them up with additional tools.
Asking an LLM for a list of every county in the USA for example, or every county with a population of more than 100,000 people.
Even if those county names and their populations are mixed up in their weights, the nature of next-token-prediction does not lend them to effectively answering comprehensive, detailed questions like that.
An agent system build on top of an LLM can do it, if it has access to tools which can help access eg a table of counties and then filter them with SQL or Pandas or similar.
amelius•Aug 14, 2026
Yes, I was assuming they'd use external tools. Using only the raw LLM doesn't sound like a good strategy.
Considering that agents are not a new concept, why isn't this a solved problem by now?
ACCount37•Aug 14, 2026
Agents are a very new concept.
We've got the early LLM-based AI agents in 2023, and it only became a popular, mainstream thing in 2025 - with Claude Code.
apwheele•Aug 14, 2026
This is another riff on not embedding a full document, but doing a summarization of the document and embedding the summary for RAG. Nice usecase for high cardinality data!
I guess it is based on the same fundamentals as well.
ipsod•Aug 14, 2026
Just this week I tried doing something similar with a nasty vibe-coded codebase I was trying to organize. I had Gemini Flash 3.6 classify each function/method in a similar way, giving a few plausible classifications for each (one agent per method).
It didn't end up being very useful - I ran a comparison where I just had a bigger agent do the organization in a more straightforward way, and that had better results.
I did find that Flash 3.6 High was >9x faster than Luna xhigh for this task, and got very similar results, though.
sheepscreek•Aug 14, 2026
I’ve read a few different accounts, including OpenAI’s own admission, that Terra Medium or higher will likely produce better results than Luna xhigh and cost about the same or less.
Majromax•Aug 14, 2026
> In the notebook, I compute a MiniLM embedding of every real Wayfair classification. I compute the embedding of the fake, hypothetical embedding from the LLM. I then dot product the fake embedding into the real ones to find the most similar. Producing: [the right answer]
Isn't this begging the question that the hallucinated classification will be more selective with respect to the real schema than the query itself? What would the dot product of <E(search query), E(schema)> have given?
Even if that is too vague, smaller LLMs are capable rerankers; return the top N matching true categories and ask for a contextual ordering.
softwaredoug•Aug 14, 2026
Yes what you're describing is a classic way of doing query understanding.
I've found, though, getting it in the language of the vocabulary has generally improved performance.
Further, when searching for "blue shoes" you want to separate the color from the item type. So its useful to have a dumb LLM do this for you. And with the LLM in the loop, its further useful to get it into the language of the taxonomy to improve embedding retrieval accuracy.
There are of course many ways to skin the cat here :)
einpoklum•Aug 14, 2026
In the past, people would post advice on how to do something clever and useful yourself. Now, people post suggestions on how to talk out the side of their mouth to coax ther magic-8-ball slop generator to say something useful.
sergiotapia•Aug 14, 2026
This is a really great trick, woah!
amitpoonia19xyz•Aug 14, 2026
This is basically HyDE (Hypothetical Document Embeddings), no? I had tried this approach in the past, worked with limited success.
memjay•Aug 14, 2026
We have this running in production. Can get pretty expensive and slow. We are trying to replace this with cheaper and faster methods that don’t hammer our LLM and elastic search endpoints as much.
Nice trick. Couldn't you embed the query though, compare it to the embedding of the categories, then ship only categories that are close to it in the prompt to a smaller model?
tantalor•Aug 14, 2026
Yeah I had the same question. What's the point of the intermediate step?
vessenes•Aug 14, 2026
Agreed that you almost certainly can just embed the original with most modern embedding models.
softwaredoug•Aug 14, 2026
Yes absolutely that's another good trick.
Even better is to search the corpus first with like naive BM25 / embedding search, aggregate over top N to get most representative categories, then have the LLM categorize in that set.
kgeist•Aug 14, 2026
It's basically a variation of HyDE (Hypothetical Document Embeddings), and the rationale is that the embedding of the query is not necessarily close to the embedding of the answer. If you generate a hallucinated answer, it can line up with the actual document better (in the embedding space, via BM25, or hybrid).
But honestly, it only works for common knowledge that's already in the LLM. If the target document contains very niche or private information, then the hallucinated answer's embedding can be even farther away than the query's.
liampulles•Aug 14, 2026
It might be a little worse, but it will definitely be way cheaper.
claudiosf1•Aug 14, 2026
Smart trick, but assumes the “dumb” llm is smart enough not to derail into an article about the lives of South American red ants. Obvious exaggeration, the point being outcomes should stay strictly within topic, avoid unrelated bloat and hit the target.
phoghed•Aug 14, 2026
If you use structured outputs they’ll usually stick to the program. Not to completely constrain the categories like TFA was saying, but something like
{ rationale, categories }
Where you don’t really care about the rationale but you’re using it as a pseudo thinking for models that don’t support it.
Luna is surprising capable and cheap, and I haven’t done this type of thing since before GPT 5 so might not be such a useful trick now
smallnix•Aug 14, 2026
Since you map each breadcrumb of the path, how do you deal with differing lengths that would be more appropriate?
iandanforth•Aug 14, 2026
No? This is just giving up and hoping.
chrisjj•Aug 14, 2026
No change from regular chatbot coding, then.
motoxpro•Aug 14, 2026
Is there a solution you are using to solve this that is more accurate and cost effective? I'm working through it now so would be curious
runarberg•Aug 14, 2026
Is scraping and putting this in a structured format too inaccurate or expensive?
motoxpro•Aug 14, 2026
That's the whole problem. If you have tons (100s of thousands or more) of labels, then you have "structured" data, but how do you correctly classify that scraped item into the correct label?
Putting all the labels into the LLM is super expensive per call when you have millions of items to classify.
You can't reduce the number of labels becasue they are correctly organizes/structured. This class of problem exists in many different domains.
runarberg•Aug 14, 2026
100s of thousand? In that case I would label about a 100 by hand and train a supervised learning model.
This problem has also been solved for 3 decades now.
arjie•Aug 14, 2026
Prompt expansion of input to extra categories makes sense if your embedding isn’t working well. But on its own, why use the LLM at all? I think you could have demonstrated the original step first and then shown that it’s useful.
Sharlin•Aug 14, 2026
I can’t believe programming is now at the stage where advice like "first have the computer give you totally wrong answers, then just find a function that maps the wrong answers to the correct ones!" is a thing.
agos•Aug 14, 2026
the trick is that it's not totally wrong to start with
addandsubtract•Aug 14, 2026
If there is no truth, there is no wrong.
speerer•Aug 14, 2026
This is so similar to human decision-making though. First I my innate experience to approximate to what I expect is right, then I map that to the truth.
It's the same for so many things:
- reading documentation (what do I expect this function to be called?)
- finding clothes in a shop (something long-sleeved and light)
- picking the fridge for dinner
- finding a book in the library...
so many analogues where I'm not coming cold to a choice.
willturman•Aug 14, 2026
I can't believe people spend their lives finding lazier ways to classify a bunch of objects that will end up heaped in dormitory dumpsters across the US next spring.
pie_flavor•Aug 14, 2026
> On two occasions I have been asked, – "Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?" ... I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.
He clearly didn't know enough about vector embeddings.
Xirdus•Aug 14, 2026
Unironically, LLM is absolutely amazing at giving you the right answers when you put in wrong input, compared to every other algorithm ever invented.
devmor•Aug 14, 2026
This been the case for a long time!
The entire problem of search is that the user has the wrong data and wants to use it to receive the correct data. That was the start, not the state we’ve ended up at - it is unironically how we got to LLMs.
philipov•Aug 14, 2026
If information is totally wrong then all you have to do is invert it to get the truth. What was it that Sherlock Holmes said? The problem ends up being that it often takes a tremendous number of counterexamples to eliminate everything that is impossible.
Worse is when you don't know whether the answers you have are totally wrong.
HarHarVeryFunny•Aug 14, 2026
Interesting technique, but even if you're getting rid of hallucinations it seems there's still no guarantee of consistent classifications. If you need to do a semantic (embedding) search anyways, then how does this really help?
kgeist•Aug 14, 2026
A common case I have is when you don't have classifications to begin with. For example, you need to find what users complain about most. I take embeddings of all records, then cluster the embeddings into semantic groups, then ask an LLM to take a random sample from each clustered group and create a classification for that group.
This method is sensitive to the thresholds (what is the maximum distance between embeddings for them to be still considered part of the same semantic group), so I run it all in an agentic loop where an agent tries different thresholds and clustering algorithms until it's satisfied with the result, plus it may deduplicate some groups.
I run it all on self-hosted hardware, so it costs nothing to leave it running for, like, a night, and as a bonus, none of the corporate data leaves the office. I think a rigid set of manually created classifications may not capture all the possible classifications that can exist. Needs a review by a human, though.
alexpotato•Aug 14, 2026
I worked on spam classification for litigation targeting in the early days of CANSPAM [0] enforcement.
We had a similar problem where you can literally millions of email that we were pretty sure came from only a limited set of bad actors.
We first started classifying emails into buckets by From, mailserver relay chains etc as that's all we had to to go on.
Over time, those buckets got linked to spammer signatures and then we narrowed down from there.
Fascinating to see this happening nowadays with LLMs.
sirnicolaz•Aug 14, 2026
I wonder how more accurate this is compared to just doing embedding similarity of the query vector and the category labels
ed•Aug 14, 2026
New embedding models support queries, so you don’t need to hallucinate a document before finding the nearest neighbor. Curious how it compares to this approach since you’d get to skip the LLM altogether.
bonoboTP•Aug 14, 2026
What does it mean that an "embedding model supports queries"? An embedding model maps text to embedding vectors. You can always perform queries with such embedding vectors against a stored set of embeddings.
cimi_•Aug 14, 2026
I did something similar 10 years ago, but instead of llms I used word2vec to calculate a embeddings of product descriptions and map those to existing categories. The LLM approach is very likely better, but I'm curious what the cost difference is.
otikik•Aug 14, 2026
I don't know the exact syntax any more, but I expect this could be solved by a single sql query that uses "inexact but close" queries and a bunch of indexes (and perhaps tags) on each category.
26 Comments
But no classification is perfect. In search in particular, you will also want to have places for manual intervention for high priority queries.
But if accuracy matters, you can't rely on embedding sort to get a closet match. With a real test set they usually don't hold up under scrutiny.
Everything in AI is like this. You get an idea, try it once or twice, "LGTM" and you ship. Then it never survives contact reality.
Embedding sort gives you a better shortlist than the whole list, but you will probably want a heavier model to vet candidates.
Additionally you could experiment with a reranker instead of an LLM or after reranking take top-3 results and then feed to LLM as input in order to reduce input token costs.
You'd think they would have solved it by now.
(Lets ignore for now that no one seems to agree to what should be the spec sheets)
Instead: use an LLM to build a large (old-school) database of products with all their specifications. The LLM can also build the schema for that database as it finds more data.
Then use an LLM to query that database based on the user's specifications (+ add some intelligence to find nice suggestions for a birthday if wanted, but I'd consider that an extra).
The idea is to make structured queries using these protocols which can be used to fetch top products matching the user needs instead of just relying on semantic search.
https://developers.openai.com/commerce/specs/file-upload/pro...
But you'll be amazed by the abundance.
I noticed yesterday when browsing on mobile that there used to be a box where I could search reviews and it got swapped with a Rufus box. I guess somebody needs to juice their engagement numbers for an investor briefing.
honestly, Amazon doesnt even need AI it just needs a better UI, more metadata for its products and to make reviews less scammy.
Asking an LLM for a list of every county in the USA for example, or every county with a population of more than 100,000 people.
Even if those county names and their populations are mixed up in their weights, the nature of next-token-prediction does not lend them to effectively answering comprehensive, detailed questions like that.
An agent system build on top of an LLM can do it, if it has access to tools which can help access eg a table of counties and then filter them with SQL or Pandas or similar.
Considering that agents are not a new concept, why isn't this a solved problem by now?
We've got the early LLM-based AI agents in 2023, and it only became a popular, mainstream thing in 2025 - with Claude Code.
https://github.com/aurelio-labs/semantic-router
I guess it is based on the same fundamentals as well.
It didn't end up being very useful - I ran a comparison where I just had a bigger agent do the organization in a more straightforward way, and that had better results.
I did find that Flash 3.6 High was >9x faster than Luna xhigh for this task, and got very similar results, though.
Isn't this begging the question that the hallucinated classification will be more selective with respect to the real schema than the query itself? What would the dot product of <E(search query), E(schema)> have given?
Even if that is too vague, smaller LLMs are capable rerankers; return the top N matching true categories and ask for a contextual ordering.
I've found, though, getting it in the language of the vocabulary has generally improved performance.
Further, when searching for "blue shoes" you want to separate the color from the item type. So its useful to have a dumb LLM do this for you. And with the LLM in the loop, its further useful to get it into the language of the taxonomy to improve embedding retrieval accuracy.
There are of course many ways to skin the cat here :)
Even better is to search the corpus first with like naive BM25 / embedding search, aggregate over top N to get most representative categories, then have the LLM categorize in that set.
But honestly, it only works for common knowledge that's already in the LLM. If the target document contains very niche or private information, then the hallucinated answer's embedding can be even farther away than the query's.
Luna is surprising capable and cheap, and I haven’t done this type of thing since before GPT 5 so might not be such a useful trick now
Putting all the labels into the LLM is super expensive per call when you have millions of items to classify.
You can't reduce the number of labels becasue they are correctly organizes/structured. This class of problem exists in many different domains.
This problem has also been solved for 3 decades now.
It's the same for so many things:
- reading documentation (what do I expect this function to be called?)
- finding clothes in a shop (something long-sleeved and light)
- picking the fridge for dinner
- finding a book in the library...
so many analogues where I'm not coming cold to a choice.
He clearly didn't know enough about vector embeddings.
The entire problem of search is that the user has the wrong data and wants to use it to receive the correct data. That was the start, not the state we’ve ended up at - it is unironically how we got to LLMs.
Worse is when you don't know whether the answers you have are totally wrong.
This method is sensitive to the thresholds (what is the maximum distance between embeddings for them to be still considered part of the same semantic group), so I run it all in an agentic loop where an agent tries different thresholds and clustering algorithms until it's satisfied with the result, plus it may deduplicate some groups.
I run it all on self-hosted hardware, so it costs nothing to leave it running for, like, a night, and as a bonus, none of the corporate data leaves the office. I think a rigid set of manually created classifications may not capture all the possible classifications that can exist. Needs a review by a human, though.
We had a similar problem where you can literally millions of email that we were pretty sure came from only a limited set of bad actors.
We first started classifying emails into buckets by From, mailserver relay chains etc as that's all we had to to go on.
Over time, those buckets got linked to spammer signatures and then we narrowed down from there.
Fascinating to see this happening nowadays with LLMs.