Matt and Ryan Have a Chat: Dishwashers, Datadog Spreadsheets, & Electric Fly Swatters

Author: Matt Williams · Watch on YouTube

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: Matt and Ryan Have a Chat: Dishwashers, Datadog Spreadsheets, & Electric Fly Swatters · Channel: Matt Williams · 48:54 · Watch on YouTube

Topics: AI agents, AI evaluation and benchmarks, AI developer tools, AI models

Summary

Matt and Ryan catch up on everyday life before discussing where they get tech news and how AI-generated content can misuse artists’ signatures. They explore DataDog’s growing product lineup, including notebooks, forms, and spreadsheets, and consider how niche AI tools and model recommendations are changing the software landscape. The conversation also touches on new and upcoming AI models, followed by show-and-tell about eye drops, an electric fly swatter, and recent reading. They wrap up with a lighthearted exchange about fruit flies and spiders.

Key points

The hosts compare their sources for tech news, including Kagi, Hacker News, The Register, and older sites they miss.
They discuss an AI-generated New Yorker-style cartoon that included an artist’s signature without the artist having drawn it.
Matt shares his difficulty coming up with a long list of life lessons, leading to a conversation about dishwasher habits and household chores.
Ryan describes how DataDog’s expanding product range includes multiple kinds of notebooks, forms, and spreadsheets, which can make the product menu harder to navigate.
They discuss how specialist companies and AI visibility may influence which products chat agents recommend.
They talk about new and upcoming AI models, including Lonk, Beam, PewDiePie’s fine-tuned model, and smaller models that could run on phones or toys.
Matt shows Japanese eye drops, while Ryan demonstrates an electric fly swatter for dealing with fruit flies.
The hosts discuss recent reading, including Brave New World and City of Brass, before ending with jokes about spiders and flies.

Chapters

Start	Chapter	Summary
00:02	Catching up and waiting for the dog groomer	Matt and Ryan greet each other and talk about their schedules, including Matt’s wait while his dog was being groomed.
02:05	Finding reliable tech news	They compare news habits and discuss Kagi, Hacker News, The Register, and other sources they have used.
09:24	AI cartoons and artists’ signatures	They consider the controversy around AI-generated New Yorker-style cartoons that used an artist’s signature.
11:33	Life lessons and dishwasher etiquette	Matt explains why he struggled to list life lessons, prompting a discussion of loading dishwashers and folding laundry.
16:14	DataDog, AI observability, and product recommendations	Ryan discusses monitoring companies and wonders how AI-driven recommendations could affect the visibility of their products.
20:55	DataDog notebooks, forms, and spreadsheets	They talk through DataDog’s different notebook products and its newer forms and sheets, along with the challenges of a crowded product lineup.
30:14	Spreadsheet tools and new AI models	The conversation moves from Row Zero’s acquisition to upcoming models, model distillation, and PewDiePie’s experiment with fine-tuning.
35:26	Small models and AI-powered toys	They discuss recently listed models and imagine small on-device AI powering toys and interactive gadgets.
39:37	Show-and-tell: eye drops and an electric fly swatter	Matt demonstrates Japanese eye drops, and Ryan unveils an electric fly swatter intended for fruit flies.
44:16	Books, fruit flies, and the sign-off	They discuss recent books, including Brave New World and City of Brass, then joke about spiders and fruit flies before ending the broadcast.

Quotes

They’re sarcastic, sardonic. — 06:15

I am also not allowed to load the dishwasher. — 13:37

We’re at like the ribbon problem now. — 29:12

I could see this going being a core part of some really cool toys. — 37:32

my understanding is spiders will take care of some flies for you as well. — 47:53

Concepts and entities

Kagi, Orion, Perplexity, Hacker News, The Register, The Atlantic, The New Yorker, ChatGPT, Datadog, Dynatrace, Arize AI, Sentry, PostHog, Row Zero, Databricks, Beam, Gemma, PewDiePie, Qwen, Ollama, Cloudflare, Raspberry Pi, Matt Williams

Full transcript (original language)

00:02 Yo, how’s it going? >> Oh, wait. I forgot to add you. Add assign to guest one. >> There we go. Howdy. >> How are you? >> Good. How are you doing? >> Not too bad. I’m at studio today. There’s some construction going on at home, but otherwise, not too bad. The sun is shining. I I am at home today. >> I see that. I see that. We’re We’re swapping spots here.

00:32 >> I I um at uh 11:00 I dropped off my dog for grooming. >> And then I hung out at the like the gas station nearby waiting for the call to say, “Okay, pick up the dog.” Um thinking it it’s usually like an hour and 15 minutes or so. Uh but at uh >> uh 11, no it was at 12. And then so at 1:30

01:03 1:30 she texted me saying, “Okay, be ready in 10 minutes.” Like, “Oh, kind of close.” >> Um but uh I got there at uh and usually I’m like 10 minutes late. So for a 10-minute warning, I’m there in 20 minutes. And this time I was there in five minutes. And it says, “I say 10 minutes, I mean 10 minutes.” And then the dog was going nuts. And so it took a little longer to do the finishing touches. And [laughter] so I got home. I had five minutes to spare. I thought, do I go to the office? Well, knowing that

01:35 we were starting now, ah, I should have just gone to the office, but oh well. >> Now I have an assumption. The assumption is when you’re at the gas station, you’re reading ebooks. Is that correct? Is that how you >> I don’t want to say waste time, but spend time. No, this time I was trying to catch up on news. >> Okay. >> Uh in preparation for our >> Yep. Fair enough. >> Fair enough. >> But uh but I am reading a lot of books. >> I’m on to a different book now.

02:05 >> That is fantastic. >> I I feel like um a lot of times what I expect I guess what I expect from a phone is to help me with those times where I don’t have a lot to do, you know, and I’m trying to waste time. So like that’s that’s what I’m thinking of like what do I what do I reach for on my phone to waste time. I do go through hacker news. There’s a couple apps on the app store that are like a little bit better. But generally like I won’t read it dayto-day but when in downtime I’ll I’ll actually go in and check it. So when you’re talking about news I wonder where where do you check your sources? Like is it I

02:36 used to use Perplexity like I I honestly had to delete the app because it it just misreports news. It’s really bad. I’m Perplexity updated to like mostly focus on their desktop app that >> it’s useless. I mean, not useless, but it’s not good for me. It’s one of the every now and then you get a reminder that um a yearly subscription is cheaper at when it when they when they promise

03:08 to have the same product but when it’s a different product then the year just ends up being well oh well that was six months down the t six months subscription down the tube I should have just gone monthly um because yeah and I I’d unsubscribe. I mean I will unsubscribe now because it’s basically worthless. >> So else do you go for news? What’s the what’s the source? Um for now well I will um for tech news I tend to use

03:39 Kaggi >> right >> um so news ki >> k a gi.com >> uh the people that make the well that have a search uh that I subscribe to >> and that um offer a browser which is just as far as I’ve seen the best of the browsers at the moment. Um, so and that’s Orion. >> Orion. >> Oh, Orion.

04:09 >> That’s me. Did you know? Uh, I did not know how to say that word until like way too late in life. Probably like 30s. >> What? >> I I would say or oron cuz it’s just like onion. But, you know, there’s in there. [laughter] the the OI. Oh, AI. I was trying to think what was the AI? Uh the AI is a uh was the was it the AI? Um the the kind of the uh the bad guys in um the later seasons of

04:42 Stargate. Uh >> oh, I had not seen that. >> Atlantis or Stargate one. >> Um and the >> I think there’s a game or >> Oh, yeah. There’s always team. Yeah. >> Uh >> I still haven’t played it. I think I own it, but I still haven’t [laughter] played it. >> I I asked because it it seems like finding a good source, like a good news aggregator is difficult these days. And I don’t want to say Reddit. Actually, I’m I was trying to not say Reddit

05:14 specifically because I used to use something called old.reddit.com, which is like how I I would browse it. And I think they’ve just removed access to it. if you haven’t been logged in to Reddit and hit that site within the last six months. So, you can’t use it anymore basically, right? >> They’ve got like a it’s aging people out of it. >> And I guess there’s not really a universal place for tech news. I mean there’s there are um I mean one of the other big sources

05:44 which isn’t very I mean it’s not very complete because it’s just >> well it’s uh uh hacker news which is Y cominator >> but yes there are plenty of people that I’ve dealt with who have never heard of uh hacker news or y cominator so >> right >> um >> I mean >> and it’s got its own it’s skewed a little Right. Obviously the commenters is skewed, but even that like the the stories that are submitted are kind of skewed. I used to read the register or

06:15 lrig. It used to be the register.co.uk. It’s.com now, which I it’s a long time out. But those ones are are good. It’s just >> they’re sarcastic, sardonic. They’re definitely there’s humor involved, which is good because it’s like different than the rest of the aggregators, but like is it funny? >> They’re funny. >> Yeah. Yeah. Yeah. Yeah. [laughter] Exactly. Is it funny? It’s it’s not it’s kind of unfunny. >> It’s not it’s not like the onion that’s trying to make fun of make make fake stories but um >> right

06:46 >> uh they will I remember the register was always a pain when I was at Microsoft because I mean they just like >> even if even if Microsoft did something good they would say oh >> it doesn’t matter [laughter] >> it’s it’s awful. It doesn’t matter. >> It doesn’t matter. >> Yeah. Speaking of bias or like a slant, editorial slant, I guess, is what we’re trying to say here. [laughter] Aler definitely has it. I’ I’ve been there a few times recently. The other thing I used to read was called the Web Weekly, which I realize is a super niche take. U but it was a website all

07:18 done in comic songs making fun of hacker news in a way that the register would. So, it was actually kind of interesting, but it’s it’s it came back during co I think 2022 and hasn’t been there since. But it’s difficult to find >> that I miss. I mean there’s I guess slash do kind of exists but it’s I mean it’s nowhere near what it used to be. >> Not the same >> 30. >> And it’s also very Linux or specifically like different types of license focus. Like it’s it’s a GPL kind of [laughter] >> for a while. Uh everything I knew came

07:50 from slash do uh fresh meat and uh company. I I don’t know that last one, but I do know Fresh Meat. [laughter] >> It was It was a um That last one was um uh a kind of spin, not spin-off, but um making fun of a magazine called Fast Company. >> Oh, I remember Fast Company. The same logo except instead of Fast, it was the

08:20 other word. Um and uh and they would always make they would point out they would always highlight um uh startups that had failed. >> Nice. >> Um so here’s who who went bankrupt this week. >> Um >> who get e fed or so to speak. Yeah. >> Yeah. Or it was a lot of Oh, what was another one? Um there was uh the one that was uh for longer articles. Um,

08:51 and it was another one of these sites that was great and then I think it got acquired by somebody and went terrible. Uh, Slate. >> Oh, yeah. Slate.com. Yeah. >> Actually had like I think they got rid of some of the writers, too. So, it was already kind of going downhill and and then >> Microsoft I think bought it out or or something like that. >> Interesting. Wow. >> Um, or maybe it was originally something with Microsoft and then they left something. I feel like Microsoft was it either the reason why they were good or the reason why they went bad.

09:24 >> I can’t remember which >> Slate was was really good. I think I’d hold Slate up to something like The Atlantic, which is like high high praise. It was just like really really well written and there’s just not a lot out there like that anymore. thinking of Atlantic or late like real news sources and also um New Yorker. I I think of Atlantic and New Yorker. Uh >> and [clears throat] I I saw uh >> I would say The Atlantic is better written than The New Yorker. And I’m I

09:55 won’t die on that hill, but I I’ll come pretty close to you. >> I think The Atlantic’s way Yeah, the New Yorker’s slipped like a lot has slipped. >> Okay. Okay. But um uh I saw it was in the news recently because or our news recently because um uh Chat GBT is generating New Yorker style cartoons which is fine. I mean they’ve been doing that. >> They have a license, don’t they? >> I don’t know. But uh part of the what

10:26 gets generated is the signature of New Yorker artists. And so apparently [laughter] there was one after Dolly Parton died that uh it was a New Yorker style cartoon with the um signature of one of the main people there. >> Um and it went really big. Uh and then the the artist like I didn’t draw this. >> Ouch. Um, you know, and he was like, you know, I’m I’m totally okay with uh all these

10:59 models using my my cartoons and um and and you know, tweaking them and using that and it’s learning that that’s fine. That’s I I have no control. I I put these out into the world, so I should expect um these models to to do that. I mean uh but but when they’re using my signature, >> yeah, it’s a bit much. [laughter] like you’re putting like a certificate of authenticity or something on it, right? You know what it when it’s not I I totally understand that doesn’t feel right at all. >> Yeah.

11:33 >> So, um >> Yeah. >> So, I wrote down I was in the office in my uh shared office uh earlier today and I wrote down this list of things to to talk about. >> That list is back in the office. >> Oh, okay. [laughter] because I was planning to be there for this. Um, and I so I have no idea. Uh, the only other bit of news I I do remember that I don’t actually need the list for was um, this came up because I

12:04 was thinking, oh, I I should I I see people who um do videos of here’s the x number of things I’ve learned in my life that I’m going to share with you. uh where x equals the number of things and also their age. And so it was my birthday this past Friday. I thought I should do one of those. Um but but then I was I like had to write down like what are the like the big learnings of my of my life that I could share. And I got to like um eight or nine and I was like

12:37 having a hard time with more. I was like I got to look at somebody else’s [laughter] list to figure out what >> you haven’t learned anything. Yeah. and and it’s like coming up with the other 47 is going to be really hard to to to uh so I don’t know if I can come up with 56 things that I’ve learned. Um >> can you load the dishwasher? This is something that at home is like a big >> ah [laughter] >> because it it’s not just loading the dishwasher because you live with somebody and like your dishes are >> exactly so [laughter] that the water can

13:07 spray on all the the correct side. Um, and so the dirty side isn’t put up right up against the wall of the dishwasher. >> Uh, yes, I can. Um, it’s interesting [laughter] that the >> there’s an emphasis on I there. Yeah. Yeah. the um uh the groomer that I use, his her her her husband uh was unloading the dishwasher and this exact topic came up and apparently he is the [clears throat] only one that is allowed

13:37 to load the dishwasher because she does [laughter] it wrong. So, it’s a common problem >> among spouses, >> right? I I am also not allowed to load the dishwasher. I just want to put that out there. So, this is why I’m asking if you know how to do it. >> You don’t do it correctly. >> Well, okay, man. I’m going to I’m going to give you a secret. And listen, I can play dumb, but like this this is maybe next level. So, this is called uh when someone asks you to break to wash the

14:09 dishes. If you break a few, they probably won’t ask you again. So, what I do is load the dishwasher incorrectly and then never get asked to load it again, which is is fantastic, actually. and I will put things in there. But yes, that generally I’m I’m not allowed to use a dishwasher. >> I I actually find it funny. I I think I understand how the dishwasher works, but there there are other >> a lot of people don’t >> other things that play. [laughter] >> Yeah. Okay. >> A lot of people don’t. >> Um >> so I thought maybe you could teach me as part of your list of 50 is is one. >> No, I it’s it’s

14:41 I I don’t like loading the dishwasher. I mean, I I I find that the sink is a good place to store uh uh dishes, used dishes. Um but I know how to put them in correctly. Um and like one of the other interesting domestic uh uh chores that I do enjoy, I mean this unlike putting things in the dishwasher, but one thing I actually do enjoy is folding laundry. >> I I like folding laundry.

15:12 >> Good at that. Yeah, I’m not bad at it either. Yeah. What a >> I I remember at one point uh we were >> Is that on the list though? Like this this got to be on the list. >> No. Well, >> all right. Write write it down because like I feel like learning to enjoy a chore is like a big skill, right? >> That could be. I mean uh I mean one of the things on my list is don’t take a job that you’re not going to enjoy. >> Yeah, that’s fair. >> Um >> small or large, I guess, in this case. Yeah. >> Yeah. Um, but but uh but I remember there we went to uh a conference early

15:42 on in Data Dog’s lifetime and we had a whole bunch of t-shirts that we needed to fold and so we all like >> there were six of us and uh each of us took like a a pile and I was always fastest. Everybody else, they remembered their gap training or whatever training at whatever store they worked at and so they did that super precisely. No, no, no. I just like whipped and got it all folded. Everything was in a nice night

16:14 column, but I could do like three shirts the time that they would do one. >> A fascinating show that we’re doing today. >> It It is. And I’m going to quickly segue to something >> people watching. >> I I know. So, let’s let’s make it a little bit of a geni. So you mentioned data dog and I just just as somebody who does competitive analysis, I’m always looking at monitoring companies in particular, I’m really interested about AI monitoring and there are a bunch out there. Some of them are are people who have competitors that have been around

16:45 for a minute. So like Service Now that’s kind of getting to the AI monitoring or probably like AI ops I guess to start and and monitoring around that based on their acquisition of light set for example is in there. Uh, I would follow Elastic just to see what they’ve been doing because I’ve done a lot with Elastic Search over the years, watching Splunk and Cisco and whatever Cisco’s behemoth is doing >> and longer down the list because I’ve used them before but like wasn’t really the experience for me is Dino Trace and again I’m not saying anything wrong about Dino Trace but when I

17:17 >> yes uh when I used them they were on prem and you would have like a Dino trace server and everything would report to it and I think at some point in time they switched to like a SAS offering as many companies do, etc., etc. No, >> I think it is in fact German because I think it says GmbH at the end of their name and I can’t remember what that means, but I believe that means it’s it’s Deutsch. Um, they recently purchased a company called Arise. Have you heard of Arise before with a zed or a Z? >> I thought it was like a CRM tool.

17:50 >> Is that Rise CRM? [laughter] >> Rise. >> I I think uh like the Horizon logo. Um, so Arise did a bunch of stuff and I I don’t want to butcher this because like I really didn’t look into it too much, but I think one of the main things that they did was evaluations. So eval sweet or what have you and data dog of course does evalu as well blah blah blah blah blah. But they came to my attention because I think a lot of times when you ask like a chat agent about email eval was in like that top 10 and data dog might have been in like the top 50 or whatever and somebody on LinkedIn sent

18:20 me a message saying like hey did you know we did this research and you’re not up there for this and I’m like it it feels like you want me to. Anyway, I didn’t interact with this person but I did read the list and do my end testing and and sure enough they were right. When it when it came to this it wouldn’t suggest data log right away. Neither here nor there but I you know just kind of want to know what’s going on. So, I looked into her eyes a little bit and it looks like Dino Trace bought them five days ago. I >> not sure if the number is Yeah. 91

18:53 sorry 915 million and most of that was uh cash and then some of it was equity and that happened or it came into effect uh five days ago and they announced it in August which I I somehow missed back in August. >> So it’s kind of interesting. They they call them AI observability leader on the Dino Trace side. I’m not sure if that’s true, but I do think that when it comes to the corpus, it it was certainly aware of a rise. It would you pull out a rise

19:23 saying like, “Hey, what should I do for this?” >> and and it would suggest a rise. So, I I feel like and and maybe again, this is just me uh pontificating here, but it feels like the fact that they’ve had that like ranking or um is it proclivity like that the agents would want to recommend Arise to You. So, it feels like it’s kind of SEO squatting at this point in time, which is weird, but that’s that’s kind of what I wanted to talk about is it feels like we’ve had a few examples of this before, like a CLI app had an argument that was a popular

19:54 hallucination, so they added the argument in in order to make sure that they could get both ends kind of thing. And it feels like one of those things. If they’re already reaching for a rise, why don’t we just buy a rise and and make that D trace from now on? So, I don’t think Dino Trace was on the list at all. like what what they do or or did at the time wasn’t comparable. I think they do have something similar now, but it it just wasn’t nonetheless. And it feels like that’s what we’re doing. Of course, at Data Dog, like listen, I I these are my opinions, not not theirs, but like data dog got a bunch of products and it feels like it’s

20:25 difficult to become a leader in all of those products when it comes to the corpus because of course some of them are are newer. They’re being developed and and being released. >> Some of them are older and and maybe have less attention on them. So I use like notebooks all the time as my agent memory just because it reads markdown files pretty easily. So like that seems to me in the corpus and it works really well to be able to put things in and out of notebooks as as a checkpointing thing. Again you could use I think obsidian be like the the de facto kind of answer but anything that kind of

20:55 works with mark since it works as an agent. >> You mentioned notebook is is the notebooks product in data log still going? Uh [laughter] again, uh my opinions, not not those of any else, but at one point in time, we had three notebooks products. Yeah, I think we’re back to one. [laughter] We’re back to one. So, we we had we had notebooks in logs. I think we could still do like log notebooks. We also had notebooks for like the the Pythonic kind of notebook. So, like imagine a notebook where you could write. And I think that one is maybe either in progress or or

21:26 what have you. And then there’s notebooks themselves, which at one point in time could do like slideshows and all all kind of stuff. the OG one’s still there. Uh the log notebooks are also still there. And I think the third product is either in preview or or or what have you, but it’s it’s more of like um what would a data scientist assume when you say notebooks? It means something different than I think what you and I would think when we were talking about notebooks or scratch pads as we’re doing. >> But yeah, def definitely still a product. >> Um >> but also like when you get three products with the same name, that could

21:56 be hard in the corpus too. When I was almost leaving Data Dog, I remember notebooks being a product that they were that was near end of life, but [gasps] more life has been blown into it. It’s >> I am I am a massive fan of notebooks because I found like it was a really easy way to write a log of something while you’re doing the investigation at the same time. So, the way I used it might have been a little bit different.

22:27 That was the way it was originally thought of. >> Yeah. And then it got a few features like being able to checkpoint graphs which was a big deal because some things can time out of. So if you’ve done an investigation a year ago and you go back to do it again and you can’t bring up the same graph and it’s yeah it’s it’s difficult to actually show it. So you could checkpoint that and I thought that was like a massive release. I was also maybe one of the few people that used the uh the screen share functionality as well. Anyways it’s difficult to develop a product. I understand. My understanding is also that notebooks is

22:59 not free but but doesn’t necessarily cost money and again I’m not an expert on data billing but I think it’s difficult when you’ve got a free product like I don’t know let’s just say Google reader for example one of my favorite examples of a a great product that people do use but is free and as a result is is not necessarily helping Google. I’m not saying that’s what’s happening with notebooks but it is a lot easier to give a product that’s making money more devs basically. Um, now in this one, it does look like I’m not alone in using notebooks as like a

23:29 checkpointing system for agents. Um, it it’s actually pretty fun. Uh, but I think that’s more of like I could write a markdown file on my own or I could share it in Google Drive or something like it’s not it’s not like a big aha moment. It’s just like all the notebooks are there and then when I go to use an agent in app like like Bits AI or what have you and Data Dog, they can also read those notebooks or or make dashboards from the notebooks. Anyway, it it’s a nice little synergy, but but just again just this to say that like there are a lot of products in something

24:01 like a mature monitoring platform like Data Dog. I’m sure Splunk has very similar things where they’ve got like a bunch of products, but it looks like the point we’re at now is that these these kind of like micro monitoring companies are are getting fame by doing like one one sliver or one thing. I think the other one that came to mind was like um oh trying to think is it not service now oh shoot my it’s for error tracking sentry is the other one that comes up a

24:31 lot whereas sentry does I I honestly I think pretty good error tracking and data of course error >> tracking do that too I was thinking the main product being uh um the podcast [laughter] >> I okay so Like I’m obviously a fan of I think Syntax is the podcast of like >> real talk and like you know they they have similar trappings as as you and I where they get off in tangents and things like that. But I I really like the production. I I think the the the

25:03 voice quality like the audio is someone who’s talking on AirPods today is is really good and often they are timely. So like a day or or two days after something so that that is difficult actually to reach. it’s hard to hit that uh that bar which I I think is great. The other one is um Post Hog. I’m not sure if you’ve heard of Post Hog and it’s sort of the error tracking uh area and like I’ve gone to conferences and talked to people about the company I work for Data Dog and they’d be like no I’ve never heard of that. Is that like Postto? And I’m like how how have you

25:34 even heard of this company? Like but they’ve got the brand recognition. And again, I mean, no offense to Post Hog, it just for me, it’s something that I had to go searching for to find, not something that Dez were talking to me about, but it looked like in in the cadre of like Genai or data science that what people are reaching for are different tools that you and I might have reached for years ago. Like I would reach for cacti obviously or or maybe I’d reach for, you know, like the the older tools.

26:04 >> I I I got a little distracted and I started looking through the documentation for data log. >> Oh, no. >> And I ran [laughter] I mean there’s it’s it’s a lot bigger than when I ran documentation. >> Yes. Yes. >> Um it’s uh there’s there’s something I I cannot figure out why data dog is doing this but there are forms >> yes >> forums. Why is why is data dog doing forums? I

26:36 mean sorry this is not Why is Joey dog doing forms? >> Okay, so I I have a I have a number of things to say here because I actually thought the same thing. [laughter] It’s like what? Um >> fired or anything, but you know, hey. >> No, no, no, no. Listen, again, I’m I’m a person. I I get it. I I thought the same thing trying to think of like where would I I use this or where would the forms land and then I realized we have a spreadsheet product. Actually, it’s exactly where it forms.

27:07 >> Yeah. Yeah. Yeah. Right. So, like when you have a sheet product, then you probably want a form to fill. >> Yeah. And and I think that’s probably where it is. But but again, haven’t used forms yet. So, let us look at the docs and confirm that when you fill out the form, it goes into a sheet, not a notebook. >> Where do I find sheets? >> I think you should just type in sheet. That should work. >> And we do have Did you notice we have AI search as well? I’m not sure [laughter] came across that as well.

27:38 I >> know. I believe we purchased a company that did spreadsheets. >> Wow. >> But you you know why we’ve got sheets. I I think it’s called uh if there’s a red team and there’s a blue team, generally there will be a purple team and the purple team deals in spreadsheets. So imagine you’ve got to go down the list of like have you are you compliant with this blah blah blah that that kind of thing. So it seems like a way to to

28:09 bridge that gap. Um I’m just trying to see I’m trying to find the acquisition. My apologies. >> It had sheen the name but I can’t I’m so sorry. >> Fascinating >> indeed. Yeah. Yeah. So, we’re we’re almost at the point now we’re not Google Docs, but do you know what I mean? We’re like notebooks are kind of like Google Docs. Sheets are kind of like Google

28:41 Sheets. We don’t have a calendar yet, but we do have a scheduling widget for on call pedag. My apologies. U now action desk was something different obviously than what sheets are in in data dog but that that’s my understanding. >> Wow. >> Yeah we we have quite a bit to the point now where the menu system is confusing if that makes sense. You know how like

29:12 when office had >> when I was still there. >> Yeah. So so think about the is it the ribbon the introduction of the ribbon. I think we’ve talked about the ribbon before but we’re we’re at like the ribbon problem now. And again this this is difficult. You got like 30 40 products PMs vying for position. Then you get users trying to use this. Yeah. It’s >> it can be difficult. >> Wow. >> Yeah. So I think the biggest one was pivot tables

29:43 back [snorts] in the day for things like telemetry logs and cloud class. No, in in data dog sheets. I think that was kind of one of the big so you’re you’re getting data in this way but you actually need to see it this way like this way >> and it’s difficult to do that when you’re not there’s no way to like manage like generally data within data dog is is written right you’re not like going back and changing the logs or anything like that but if the view needs to be that way it’s difficult to rationalize through like a dashboard or a notebook or yeah sorry this is a lot about data dog

30:14 but yes we’ve added quite a few products I would love to give you a tour sometime Yeah, >> I I feel like I need a a new to updated tool. >> Um well, I thinking of uh spreadsheets. Uh remember I did a video six months, eight months ago about uh a tool called row zero, which I thought was pretty cool, >> right? >> They got acquired by data bricks. >> No way. >> Yeah, >> they had in fact. >> I I don’t know.

30:45 >> Wow. I don’t know if they were hurting for cash and needed to >> needed [clears throat] to do something to survive. >> So, one thing that I’ve heard commonly is that like prompt management and evals are typically done in spreadsheets just in terms like just me trying to think of like why data bricks would be interested in that. Yeah. >> Yeah. I mean, if I needed to store I mean, prompts often you just need somewhere to store it and it’s a I mean, it’s a convenient place to store that

31:16 kind of stuff. Yeah. >> Yeah. Absolutely. So, you can imagine like the cell doesn’t get any bigger. You paste in the prompt so it doesn’t take up much space and then the rows can be like, “Yeah, I tried this with this agent, that agent, this model, that model.” >> Um, unrelated but in the same industry, did you hear about Lonk? Uh, [laughter] >> yeah. I know. When I pulled up the register earlier, it was the front page, which is why it kind of joged my memory, but it’s it’s like a a multi- trillion uh parameter model. I don’t think it’s out yet. >> One trillion. Okay. And it’s not

31:47 released yet. I think it’s going to be on Hugging Face next week, but I’m I’m here for ML. I think I’ve said this >> and I thought it was like the 24th. >> Oh. Oh, in a few weeks then. That’s >> Yeah. >> Ah, that’s too bad. but maybe two and a half weeks. Um, but uh yeah, and I thought one of the big things wouldn’t uh that they were able to train it on fewer GPUs than previous models, more efficient training process or there

32:19 were a couple >> there was that they said they used 4000 so um >> uh something like beam or bead or something like that. another model >> beam. Yeah. Which which kind of took me off because there was also news about the beam [laughter] up there. But anyway, I realiz and Gleam and I I understand that no one cares about Gleam and and the beam and Erlang, but yes, Beam the model. I know. I see you laughing. [laughter] Uh Beam the model. I I didn’t do too much research into because the name collision kind of screwed me over.

32:52 >> Uh there’s Beam and then there’s also um PewDiePie’s model. Have you heard of this? >> I saw him showing up in a thumbnail, but I didn’t really see what was going on. >> Okay, so a couple things. Um, PewDiePie wanted to make a model, and he did. I think he forked Quinn uh 3.59 billion, maybe 3.69 billion. Um, what he wanted was a model that would work with Odicius, his harness, better than just offthe-shelf models. So he wanted to

33:23 fine-tune basically a model to work with Odysius better. >> Um so he made a model. >> Mhm. >> It was terrible. >> He fine-tuned it. [laughter] And I don’t think it got worse when he fetuned it, but it it wasn’t necessarily better. It was like maybe a lateral mood >> uh move. And then he’s like, “Hm, I’ve done this before. Maybe I should distill it.” [laughter] >> He’s like, you know, not that I’m going to, but if I were going to distill it, what would I distill from? He’s like, I really like this soul model. Maybe maybe I could

33:54 figure out the thinking um tokens cuz apparently it encrypts them or or does them like off by one. It does something interesting to like encrypt the token so you can’t see them even though your agents getting them. So it it reads them and I think that’s how you get into it. As it’s reading them, you can kind of like piler them off. Uh, not going to say he was distilling because he kind of took that approach in this video, but let’s just say he got banned once and then said, “No, I wasn’t doing that.” And then maybe got banned twice. [laughter]

34:25 >> And and like I’m sure, you know, some legal team is is going to say something to him. Uh, in his defense, the the aside he just happened to be talking about when he was talking about legally is like, “Hey, if you can steal all of humanity’s data, it feels like that’s sort of my data, too. >> [laughter] >> It’s like this is very slippery slope reasoning. >> Yeah. Yeah. >> But I could kind of see where he got to. Anyway, don’t don’t distill models and and lose your accounts. But uh it it feels like he may or may not have done so. >> Argument used by I don’t know one of the

34:55 Chinese models that uh right like well you stole the information from everybody else. Why can’t we steal from you was the was the gist of their >> it and listen not for nothing but I do understand that stealing data is one thing and then like massaging it and making it work and training your model is a whole other problem. >> Uh I but yeah I mean you can get archives of Wikipedia and stuff if you want to train on these that’s not that difficult to do. >> I think distillation is the legality of distillation I think is it’s it’s not

35:26 whether or not it’s legal. I think it’s against their terms of service which is what they hit them with obviously >> but yeah beam Ajax and uh Lon are the ones on my >> and then did we get did we unmask the alpha model was it something alpha >> huh >> what was the name of that that that model space something alpha >> the bunny one >> space bunny alpha space bunny alpha

35:59 that we talked about last week. >> Still, >> yeah, it it still hasn’t been on mass. We still don’t know who who made it. >> Yeah, it’s nothing. Sorry, I thought we might >> I see there is um I was looking at models listing under newest. Um, so there’s a new there hasn’t been a a new embedding model on Olama for a while, but it looks like there’s embedding Gemma too. >> Um, >> right.

36:30 >> And uh, and then there were the two uh, Cloudflare models, KFT Cle. >> You heard of Cliff? >> I heard of but have not used. Interesting. >> And then Lia or Leia. And again, Lia is different because it’s BERT based, right? >> Yeah. >> We’re going back to BERT, modern BERT large, [laughter] >> which is how the all the um classification models were before before the Jev stuff happened.

37:02 >> Yeah. And then the one right underneath that is Tev. Have you played with Tev yet? >> Yeah, Tev and Nimble. >> Uh and Tev is interesting because it has a point. 0.8B model. for really small machines. >> Um, >> yeah. >> I I have a feeling like like for phones that could be a big deal. >> Yeah. Yeah. >> Like I I just based on most of the examples I saw for Jeb, it’s like I

37:32 could make something to run a free-to-play game on my phone to like collect my daily re, you know, this where my brain went to automatically. I could see this going uh being a core part of some really cool toys. You know, something you could load up on a >> Yeah. >> pie or something even smaller. >> Um and now it’s controlling a car that is actually smart to >> remote control, but you just kind of point it in the right direction that goes around obstacles. That’d be awesome

38:04 toy. >> I was thinking about like Furbies or Magic Eightball. Did you have a magic eightball before? Yeah. like with a mic on it or something. >> Yeah. >> My my daughter just uh at the uh farmers market on Bainbridge Island, there’s a guy who uh has a demonstration on on bees and he’s got like a little beehive there. >> Um and uh and my daughter always goes to that uh booth and she is like looking for where’s the queen today. Um, and next to completely unrelated, but right

38:36 next to his uh bee display, he’s got a magic eightball. And and my my daughter picks it up and says, “What’s this?” Like, “I’ve not told you about a magic eightball. What? How?” >> Yeah. [laughter] What? >> It says, “Well, you just ask a question. It tells you, how does it work?” Well, shake it up. And show she got the tutorial on how to use a magic eightball. It’s like she did not want to leave that area of the farmers market for a while. like, “Oh, let me just keep asking questions.” >> And I said, “We have one at home.” And

39:06 it’s it’s >> somewhere above right here boxes. And I need to go through and find that box that has my magic eightball cuz I still have the one from, >> you know, in when I was >> whatever age I would have been 10 years old. Um, >> it feels like it’s fast enough to run on device that it could give you like a magic a-ball answer >> based on you talking to it and doing text to speech or or speech to text. >> Yeah, >> which is Yeah, it’s amazing.

39:37 >> Awesome. >> What a time to be alive. Um, did you have any other news that we should I did have one check out item, but it’s not that interesting. >> I have a checkout that’s gonna be more of an unboxing. >> Oh, mine could be an unboxing, too. Fantastic. All right. >> Well, >> it won’t take very long. >> Sure. >> You go first. [laughter] >> Yeah. So, what I’m showing you are uh as I drop them. Uh roto drops for contact lenses. No, this is uh

40:08 eye drops. >> Oh, okay. >> And then we’ll we will unbox it. Oh. Oh, look. This is how it opens. Oh. Oh. Oh. This is what Japanese eye drops look like. They come in a little sealed thing. I won’t do it on the uh on air because they’re very cold. >> This is I ran out of the other ones. They’re quite cold, but they are very very cold and they work on on contacts, what have you. Um but this is like the be like the beginning of of the Japanese cold eye drop level because they get there’s like a special special Dragon

40:39 Ball slime version. That’s just ridiculous. for example, I I know >> and things are different in Japan, but these roto eye drops I think are really good for me because I I end up doing a lot of work where I’m just reading text, you know, these days I’m I’m just reading for for hours reading and generally I’m hitting cancel or hour or what have you, but it’s a lot of reading and I find my eyes dry out pretty quickly, especially this season where it’s kind of like >> hot and cold. So, I might be using the AC or a fan or the heater or what have you and it feels like it dries up my eyes even more. But that’s I I think for your health, if you find yourself

41:10 staring at your prompt screens or or the uh I don’t know the reasoning tokens scrolling by, we can do some good eye drops. I like >> Okay, so something I have not tried yet, >> but I’ve already taken it out of the box. Um, and uh I saw somebody last night I was walking my dog and uh as I was walking my dog uh somebody was somebody I uh ran into somebody I knew

41:40 and they were returning something to a ne a neighbor that they had borrowed and I saw it under their arm and I thought, “Oh, is that just like a um a pickle ball like a pickle ball racket or something?” because Banebridge Island is the where pickle ball was founded. >> Um >> I didn’t know that somehow. Sorry. >> Oh yeah. Pickle was founded here on Bainbridge Island. The first pickle ball court was here. Um and uh but no it was not. Um so at the moment everybody on

42:13 the island and probably in other places around this area is being inundated with fruit flies. Fruit flies [laughter] are everywhere. And this thing is a electric fly swatter. And so you can just kind of swat at the fruit flies and they go >> as they go get fried up in the in the little net netting that’s in here. So >> uh I think I can Oh, there’s a tape on the

42:44 >> um button it on. >> Okay. And I I I don’t want to electrocute myself, >> but it would be if I >> Yeah. Stick your tongue in there. [laughter] >> Oh, it’s on. >> I think you you you should feel what the the flies feel. You know, you’re bigger. It shouldn’t shouldn’t hurt as much. >> I don’t see anything to electrocute. I mean, uh zap. >> It’s like a modern torture device.

43:14 >> It is. It is. Um, [laughter] so I’m very excited about trying it. And uh, it looks like it’s delivered with a full charge. And then there’s like a another piece that feels super cheap that is a base for it. So you can leave it >> Oh, so it can charge. >> Well, yes. And you can also leave it standing in case the flies decide to fly through that area. >> Fly through it. >> Right. Um, their last flight. Uh so

43:45 [laughter] uh yeah, so I’m very excited to try this out. I have not tried it. Um it’s it does not feel like it’s the highest quality device. >> It will not survive a drop, >> but >> right >> um for uh this was like 10 bucks, 15 20 bucks on Amazon. Uh Sean covered the problem that I currently have [laughter] and that happened to you. I’m not sure if you recall that episode.

44:16 >> Oh [laughter] my god, I will never forget it. >> Yeah, >> that was [clears throat] a great episode. Classic. You should get a short of that. That was [laughter] >> That was a good episode. >> So So yeah, that was uh that’s that’s my thing. Uh the the only other um and and of course it also comes from you know a well-known brand everybody I mean this is just rolls off the tongue. >> Yep. >> Is uh

44:48 >> interesting. Yes. Yistic. >> Yes. Victor. I think there were like a um a bunch of other brands that were all the same thing >> and they all have >> but just >> random consonants and vowels put together. Um, that’s >> So, the only other thing I got is I just I finished uh like last week I finished off um uh Adventures of

45:18 something uh Adventures of uh God, I forgotten the name of that book. um >> Killian Murphy the adventures of >> uh something u which was a really fun uh fun book I just couldn’t put down of course >> ami a safari >> yes that’s it thank you >> the one we talked about last time okay >> yes yes um and that was such a great great story uh and and I just couldn’t

45:48 put it down and so then I like okay let me let me get something shorter and so I started reading Brave New World all >> Yeah. Yeah. >> Um >> it doesn’t hold up that great these days, but yeah, the Soma part probably does. >> It’s I mean for now I mean at reading it now I was like oh well this is kind of a predictable story only because it’s all I mean so many other stories have been based on it. >> It’s it’s no longer precient. Yeah. >> Yeah. Um so I I finished that off. Uh and then there was something else. I

46:19 started reading orbital by Samantha Harvey. That didn’t really grab me. But um I moved on to City of Brass, which is essay Chuck Herbat’s uh I think her first big book >> and it’s like a series of four books. Um and it’s >> Dave of a Bad Trilogy. >> Yeah. >> Yeah. >> Chuck Robbati is apparently a phenomenal uh writer and I guess you saying that the other book was fantastic kind of lends credence to that. Yeah. Um, so I’m excited about uh the first

46:52 few chapters were like the the same thing with the u the previous book I read by her. The first like two or three chapters like oh god this is boring. I’m going to skip this book but then it gets gets really good and the first few chapters were kind of hard to go through but >> this [clears throat] is great. >> Yeah. It says it’s it’s um it’s like a jin city. I think jin means genie. >> Yeah. like DJ I N Gins are like gins are weird creatures that can be like positive or negative if I understand the

47:22 mythology. >> Yeah. Whereas a genie is generally like trickster I think in our mythology. >> Right. >> Um so it’s the idea of and she’s finding out who she really the the the human in the the main of the story is is um just finding out who she might be and how she is related to the jin. >> Very cool. So, it’s pretty cool. That’s all I got. >> I mean, listen, we could always end early. >> I’m excited to [laughter] start whacking some fruit flies.

47:53 >> Oh, the the Buddhist in you is strong. [laughter] >> Um, >> all right. Well, we can call it if you want. Um, it’s, you know, to keep people around. We had some comments. >> Questions. >> Thank you, Sean. Definitely check out these new models we showed. Yeah. Don’t Don’t get hit by a spider. >> I I would say close your mouth when you’re looking back and up. >> Yeah, I know. [laughter] >> There’s no spiders floating down, so that’s good. >> That is very good.

48:23 >> I see one over there. >> Oh, shoot. >> No, he’s >> No. >> See, the other the other thing is, Matt, and listen, I know you got a new toy, but like my understanding is spiders will take care of some flies for you as well. Yeah. >> So, maybe if you were nicer to the spiders, you’d have less problems with the flies. I don’t know. >> Well, no. Oh, I think I’ve been really nice to the spiders in this room and they just torment. >> Too nice. >> Too nice. >> Yeah. >> All right. Should we find that end broadcast button? >> Find it. [snorts] Finish.

48:54 Uh end