Daniel Svonava
@svonava
Followers
2K
Following
92K
Media
538
Statuses
5K
Vector Compute @superlinked - tweets about information retrieval | xYouTube
San Francisco, California
Joined September 2012
Compare 36+ Vector Databases on features, performance and price ⚖️
20
59
302
Eval Driven Development:
Sharing an interesting recent conversation on AI's impact on the economy. AI has been compared to various historical precedents: electricity, industrial revolution, etc., I think the strongest analogy is that of AI as a new computing paradigm (Software 2.0) because both are
0
0
1
1. This will happen to all domains. 2. It will happen to narrow niches in all those domains and will be driven by thousands of companies, not just a few pioneers. Most people intuitively get 1 but almost nobody gets 2 and its implications.
I took delivery of a beautiful new shiny HW4 Tesla Model X today, so I immediately took it out for an FSD test drive, a bit like I used to do almost daily for 5 years. Basically... I'm amazed - it drives really, really well, smooth, confident, noticeably better than what I'm used
0
0
0
Bullish or bust with @basetenco - deepseek/qwen/kimi enterprise adoption still slow due to obvious geop reasons.
0
0
3
we should track the ratio of tokens of code generated vs tokens of code put into production... my claude code must be doing something like 100:1 ^_^
0
0
2
What a delightful application of my favorite technology ❤️
Artists would sometimes paint over old artworks to save money on canvases. That's why we hid an easter egg on the homepage of https://t.co/HjH80CEawH where you can 'scrape away' the background image to find another one behind it. Someone just texted me that they found it 😅
0
1
3
my LLM is frustrated that I don't get the "super clear" visualization it just made... how is your morning going?
1
0
0
Interested in building the next layer of AI infrastructure? Join our hackathon on semantic modeling to help AI make sense of structured + unstructured data. Hosted by @HamelHusain and @BEBischof. 📅 Sat., Nov. 15 ⏰ 11AM - 8PM 📍 San Fransisco https://t.co/pjcHeaK111
luma.com
Most hackathons are demo days. This is not a demo day. This is a competition of pure engineering quality — with prizes (the latest NVIDIA hardware, including…
3
6
27
“context anxiety leads to reduced reasoning” but RAG is dead… ☠️
1
0
1
im so tired of the 30-page ai-slop brain-rot around RAG here - i solved it for you. this simple graphic tells you everything you need to know.
I wrote 6 months ago that RAG might be dead. That was after an aha moment with Gemini’s 1M context window - running 200+ page docs through it and being impressed with the accuracy. But I didn’t have skin in the game. @nicbstme does. He just published an excellent piece: "The
36
77
934
If your entire data scope is one document, even with 100s of pages, then maybe you don’t need any kind of indexing of the information.. duh? By the way, the reasoning capabilities of the latest models have been shown to be inversely proportional to the length of the input.
I wrote 6 months ago that RAG might be dead. That was after an aha moment with Gemini’s 1M context window - running 200+ page docs through it and being impressed with the accuracy. But I didn’t have skin in the game. @nicbstme does. He just published an excellent piece: "The
1
0
2
I take it back, gpt5 "extended thinking" is the best solution for working on longer documents, reconciling 10+ feedback bullet points, throwing in voice mode for brainstorming on sub-problems, referencing other long documents for alignment etc. I have super-pro-max on
0
0
2
What stuck out from my session with @svonava (@superlinked) on embedding models and representation learning
1
2
7