Latent Space / episode intelligence
Speakers in the public record
Claim mix
Evidence policy
Every row below preserves an exact excerpt. Identified speakers are linked; unresolved voices are labeled and excluded from people counts.
Claim ledger
19 published records
01 / belief
“Yeah, sorry. So we filtered Common Crawl just by the top, I think, 10,000, just to limit this, because obviously there's this massive long tail of small sites that are really cool, actually.”
02 / belief
“Yeah. Because I think you, you would use find differently than people would by themselves.”
03 / belief
“I've been talking to Harrison actually about like a more structured way perhaps within Linkchain to like do evals. Because I think that's a massive problem.”
04 / uncertainty
“Like that is one title, but I don't know if this really exists because I feel like we're too rigid about like bucketing people into categories.”
05 / belief
“My take is that it's much easier having been on both sides of that coin now, it's much easier to stay obsessed every single day when the genesis of your startup is something that really spoke to you in an incredibly meaningful way beyond just being some insight that you've noticed.”
06 / disagreement
“The podcast with Amman was great, but somewhere where I disagree with him is that you need to own the IDE.”
07 / uncertainty
“If I don't know anything, I'll start by just trying to get a mental model of what is happening.”
08 / belief
“I think solving hallucinations, being able to guarantee that the answer will be correct is I think super interesting.”
09 / belief
“I've seen Phind blow up this year, mostly, I think since your launch in Feb and V2 and then your Hacker News posts.”
10 / prediction
“Because like as soon as like we told him like, hey, like we think that the future of search is answers, not links.”
11 / commitment
“That is a strategic shift, right? That you already decided to make by the time you met Ron, which is we are going to have our own hardware.”
12 / preference
“And we start working with Nvidia, which is great. And something that I love about Nvidia, by the way, is that after that intro, we got matched with like a dedicated team.”
13 / recommendation
“I think we might want to give people an impression about like type of traffic that you have, because when you present it with a text box, you could type in anything.”
14 / evaluation
“In fact, it's a little challenging sometimes to like finish kind of like the rest of like the description of your pitch because like, he'll just like asking all these questions about how it works.”
15 / evaluation
“Even GPT, like the text DaVinci 2 was available at the time, wasn't that good at generating code and it would generate like very, very short, very incomplete code snippets. And so we launched that last summer, got some traction, but really like we were only doing like, I don't know, maybe like 10,000 searches a day.”
16 / commitment
“That's what we've been doing. So we launched the very first version of Find in its current incarnation after like the previous demo connected to our own index.”
17 / evaluation
“When I returned to it in fall of 2021, when BigScience released T0, when BigScience released the T0 models, that was a massive jump in the reasoning ability of the model. And it was better at reasoning, it was better at summarization, it was still a glorified summarizer basically.”
18 / commitment
“I think, okay, let's just start with the name for now and then we can do the full Paul Graham story later.”
19 / recommendation
“I used the standard BERT and also Longformer, which came out around the same time. And Longformer was interesting because it had a much bigger context window than those models at the time, like BERT, all of the first gen encoder only models, they only had a context window of 512 tokens and it's fixed. There's none of this alibi or ROPE that we have now where we can basically massage it to be longer. They're fixed, 512 absolute encodings. Longformer at the time was the only way that you can fit, say, like a sequence length or ask a question about like 4,000 tokens worth of text.”