bobvanluijt

Somewhat related: a book that really helped me is Picturing Quantum Processes by Bob Coecke. It's a visual representation of quantum processes, which really helped me understand the concept

> issue with a founder

That would be me

"Kamer 306 is a 2006 documentary about the birth of electronic music inside the Philips Natuurkundig Laboratorium (NatLab) in Eindhoven. Set in the legendary Room 306 of the lab’s Acoustics Department, the film traces how Philips technicians, engineers, and composers opened up a completely new sonic world in the late 1950s."

Besides pure vector search there is hybrid (BM25 + vector) but now there is also agentic search (where the agent can set additional filters, do multiple roundtrips, etc on top of hybrid vector search.

Meet Elysia - a decision tree based agentic system that dynamically displays data, learns from user feedback, and chunks documents on-demand.

Most AI chatbots are stuck in a text-in, text-out world. But what if your AI could dynamically decide not just what to say, but ? Elysia is designed to completely rethink how we interact with data through AI.

From: https://x.com/weaviate_io/status/1955283683717369970

Currently still a demo video but they are looking for testers

Ha, very cool

It depends on additional filters and whether you want to use vector search only. The upside of using Faiss would be storing the ID as file metadata and embedding it in the Faiss index. However, if you need any other filters or data, you would need to store it somewhere else.

How nice to see Stockhausen on the front page. It’s definitely an acquired taste, but if you take the time to dive in, you will appreciate it

Did you see this? https://arxiv.org/pdf/2502.00032

At Weaviate we are hiring, mostly remote: https://careers.weaviate.io/#jobs

> We could make programs _so much_ better than they currently are, it’s not even funny.

IMHO, that's what the author is arguing. Now that we have the tech to do it, this is how companies will win new users. Making the technology "better" means subjective design decisions that can be defined as tasteful when done well.

Agreed, beautiful indeed

I've been working on Weaviate for 5 years, and today, I'm in the fortunate position to be able to help other founders, too.

In the process, I've noticed that many founders ask me the same questions so I've decided to answer them in an IMHO section on my website.

I would love to get some feedback and/or opinions if this is helpful to other founders, builders, etc.

Hi all – We are releasing new storage tiers in Weaviate to move data between hot, warm, and cold tiers. Both available open-source and enterprise cloud

The author gives the example of the Köln Concert; we can buy scores to play the piece, but the score doesn't contain enough information to _really_ replicate the performance. That's what I meant by the missing context.

Interesting that Nelson Goodmanis is mentioned. IMO music notation has the same, albeit lesser observable, problem as language and that's context. Regardless if it’s the Köln concert or a modular synth. If we try to capture all context for -let's say- the first second of a musical piece, it’s almost impossible to do.

> I love Noam Chomsky so much

+1

We are hiring (mostly remote) various engineers at Weaviate (open-source vector database).

More info: https://careers.weaviate.io/#jobs

Same! That was a sad day

Yes, personally, I'm actually very bullish on CPU inference for both models and the vector DB.

Demo with Google AI Studio: https://twitter.com/bobvanluijt/status/1758185143116730875

Soon more...

I have access and will share some learnings soon

“We estimate the supply-side value of widely-used OSS is $4.15 billion, but that the demand-side value is much larger at $8.8 trillion. We find that firms would need to spend 3.5 times more on software than they currently do if OSS did not exist.”

Oh, Jordan’s posts are so good. As somebody working on a database I couldn't agree more with this. Especially: https://hannes.muehleisen.org/publications/DBTEST2018-perfor...

Modules are optional, so if you want to create an end-to-end RAG pipeline, you can do this from the database directly. It's basically less abstraction based on how the user wants to feed data into the prompt of the model.

(I’m affiliated with Weaviate) You might want to check out this getting started guide. It takes a couple of minutes, and you're good to go https://weaviate.io/developers/weaviate/quickstart

This will be solved in Weaviate https://github.com/weaviate/weaviate/issues/2424

> The Auto API allows you to simply tell Weaviate what kind of search you want and WeaviateGraphQLGorilla formats the request under the hood.