20 comments

[ 3.0 ms ] story [ 57.6 ms ] thread
What does “exclusive” mean?

“Exclusive AI products…”

Can't honestly take 'vector search' as being AI.
Search has always been machine learning, so it's not a stretch to call it artificial intelligence, albeit of a limited order.
Both machine learning and AI are sort of nebulous marketing concepts more than anything else. Expert systems have also been marketed as AI.
This isn't true for machine learning which has a very specific and widely accepted meaning.
Sure, but GP objected to the term AI.

Although ML is also aggressively co-opted by marketing, to the point where it's questionable what it means in practice.

Would you say that the very specific and widely accepted meaning of "machine learning" includes nineteenth-century linear regression?
Did linear regression begin to work differently in the twenty first century?
No. Would you say that adding a least-squares fit to an Excel chart is "machine learning" according to that very specific and widely accepted meaning of "machine learning"? Is it machine learning if you do by hand on paper?
Big difference in the 21st century is we have bigger computers.
Search itself is classified as information retrieval.
It's not AI but it's definitely a tool to build AI systems.

Similarly, GPUs or CUDA aren't AI in themself but are fundamental in the day to day work of AI practitioners.

I feel the landing page needs a minimum example of how to get value, not only how to install it. Think stripe, that to this day shows how to integrate it with a few lines of code. When looking at the tutorial, I also feel a lot of steps necessary to check it out. Why not provide basic functionality out of the box or a quick start to git clone?

On another note, having built startup search myself [1], I think the demo does not do the technology justice [2]. People will most likely judge the product based on it, but the offered startup search does not work as expected, e.g. when trying "food delivery". Feel free to get in touch, will happily provide some data :)

[1] https://startupradar.co

[2] https://demo.qdrant.tech/

Thanks for the insights, really appreciate them! I work at Qdrant, and if there is any way to ease the start with using it, then we're eager to think about it. However, it's way easier to do it with a tool like Stripe, which essentially simplifies one thing. Qdrant should be rather compared to databases like Postgres or MongoDB, but designed to support quick vector search. There is plenty of possibilities of how to use it, including semantic search, image search, anomaly detection, etc. so we don't want to bring a single topic as if it was the main use case. That's why we rather focus on a variety of tutorials and demos, covering different domains.

However, we can surely think of an example based on your data, let's talk! :)

I think the "value" is explained just below: "Make the most of your Unstructured Data Qdrant is a vector similarity engine. It deploys as an API service providing a search for the nearest high-dimensional vectors. With Qdrant, embeddings or neural network encoders can be turned into full-fledged applications for matching, searching, recommending..." However, I agree that the learning materials could be organized better.
You say that your semantic search works best with short texts? What exactly do you define as a short text? Just single sentences, paragraphes or even a couple of pages?

Other than that, I agree with the other commenter that more complex demos would be great. For the semantic text search one for example, I guess it is limited to only return five results at most? Makes it difficult to judge, how many revelant entries it really found.

Anyone know how this differs from Pinecone?
Pinecone is not open-source but a cloud solution only. Qdrant offers both: an open-source solution to run locally or to deploy in your own environment, and also a managed cloud platform.
Was looking forward to a search engine for vector images.
We use qdrant. It's well written software. I do wish these vector indices would stop trying to be complete databases though.

Probably the best outcome is for large DB companies to acquihire and add vector queries as a feature.