Shreeda takeover, AI News, and three guests

all right i guess we're live this thing on it's on yeah you see the live thing on the top left oh yep yep yeah we're live at Oh it's only 720p can we upgrade that to like 1080 i want to be in higher resolution or is that all we The only option we have no no there it says high definition is 720 i also can't Oh it says quality

selection disabled while live oh okay let's see if you're watching this on your like iMac screens it's not going to it's not going to look amazing do we want to I don't think we should restart it i think we just got to go we got to run with it no yeah okay let's Let's go let's go this is for next time

okay cool okay so um if you're joining in I guess we only have three people four people wow people are are joining pretty rapidly it's skyrocketing skyrocket uh we're waiting for the fifth okay we got five if you're one of the five or six or whatever number we get to by the time I finish for every new person we're going to announce them oh

my gosh can you actually Can we actually see the participants Daniel uh so no you should let us know comment with where you're joining from but if yeah if you're joining uh maybe you're expecting Shane or Abby but instead you have Shreda and Daniel we are members of the Monster team uh I believe Shane is

enjoying life and hopefully Abby is getting some sleep being in the EU time zone right now um we are taking over I mean it's only 900 p.m i hope he's not sleeping yet he's probably not yeah hopefully he's having like a like a beer or something like that to wind down or maybe he's petting sheep at a farm or something we don't We mean you can do

both you can do both exactly um but yeah welcome to the AI agents happy hour uh Daniel how does it feel to be on uh wow the it it feels like so many things i don't even know which which emotion to begin with sure sure that's okay but yeah how about you tell it yeah go how How about you tell people who you are what do you do at Moira uh yeah um

uh Daniel as was said before uh I'm hanging out uh like an hour west of Toronto in a city called Waterlue um and I've been an engineer at MRA for like a month now I guess four or five weeks time flies um it seems both like it was very recently and that I've been here for a very long time um but before that I was at Netifi and Gatsby okay and that's how

you know yeah so I've I've I've known all the all the folks for quite a while now oh Abby says Daniel is the goat that's awesome yeah that's why he's on the live stream that's why I picked him to be on the live stream today with me um yeah i guess I'll say I'll talk about myself i'm Shida i joined MRA the end of April I believe um been kind of working

freelance for them for a bit as well uh I'm a writer can't believe I'm one of the few writers who doesn't hate AI um actually you know I don't know my feelings about AI are evolving as I imagine most people's feelings are still evolving um what is like as a writer what is like something you that still annoys you about because like you you use AI to like help you write and I

think that's like pretty common i I hate the M dash discourse because uh people are like remove M dashes it's like a giveaway that like AI has you know written something but like any real writer like most writers love the M dash i mean if you're an M dash hater I don't know you can comment why or something or

come out and say it but uh but yeah you have to out yourself don't worry they're going to have to pry it from my hands i'm sorry i love it that's the thing like I don't think anybody like I had never heard of the M Dash before and I feel like it's a writer thing that people just know about and it's like if you're not a writer you probably just like never noticed it or like it's one

of those like invisible things true so true um okay ward just likes Daniel awesome okay obby and Ward seem to be awake right now that's cool um just hanging out um okay well we're actually supposed to be talking about some AI news for these first few minutes um Daniel where should we begin oh um I mean why don't why don't we begin

at home uh we launched uh a master course today and it's a it's a pretty neat thing that we've uh put i don't know if you have if you want to like share your screen and just them it might be uh let's see if I can do it um Oh my gosh I'm I'm a little nervous but Okay let's Can you guys see this this cursor screen yeah yeah it looks pretty good okay cool

so I I just started like a new basic monster project i literally just went through the installation of Maestra and now in cursor I'm going to ask my agent to start the master course okay it's asking for my email so here we go guys for some context one of the the things that's uh neat about this so we have as many of you know or all of you know we have uh

an MCP doc server uh just to kind of help out in your IDE to be make like I don't know your cursor winds surf like an expert with MRA but we've added new tools to the doc server and the tools uh basically help you uh yeah z's asking to zoom in to oh just make the the text a bit bigger is that bigger is that good yeah that that's good and

then uh so these tools kind of like help your IDE interact with this course that we've built and so that's what we're we're seeing right now uh that were so so I guess Diego what I'll say is that the course it's like not YouTube videos it's not even like tutorials that you read I think it is the first or one of the

first courses that you literally just take in your uh code editor so like here we're just using the um cursor is using the uh MRA MCP doc server and we've built a course right here so you can just chat to your agent it's going to go literally through the course teach you about like what an agent is um it's asking me if I'm ready for the next step

and then you can see that I think this is this is a workflow right that we built in the back end like we're going through workflow steps but seems like uh if I just say I'm ready it's telling me like this is the next course material that I should learn about um I think there are some like interactive elements as well like it's making sure that I've actually installed MRA walking me

through it super super cool i believe uh Shane and Kenny on our team have spent a lot of time on this so shout out to them uh pretty exciting and then you know show me how much progress I've made on the course is what I'll type and it's super cool because you can literally just track your progress you can click this link and it's actually going to um

I don't know if you can see my browser yeah we can't see you can only see your your ID one second think you have to share your other window yeah yeah yeah no worries um I I followed that link and I can just literally see uh and track my progress super super cool like this is like a custom URL I believe to uh probably the email address that I signed

up for the course with um if you Sorry go ahead go ahead Daniel i was going to say like if you if you do check it out uh give us some feedback we'd love this is kind of like it wasn't really anything we could model this off of so there may be some things we get get wrong with it or some things that don't work so well or some things that you'd like to see added so we just love

to to get any any feedback around it uh or even like what what you want to see uh like lessons in there like maybe specific lessons or courses that are missing that you'd like to learn about as well yes um and in fact I think Shane you know in the spirit of Shane uh he would probably want us to do a giveaway to celebrate he's a big giver that's that's

what he's giver he's an educator and a giver he's anything you want him to be he's the He's the St nicholas of uh Mustra exactly um uh but yeah he would want us to do a giveaway um so I guess the first uh five people who you know respond I think the Monster Twitter account retweeted Shane's announcement um of the

course i guess the first five people who respond to that and tell us what they'd like to see uh we'll reach out we'll DM you get your address and uh ship you a copy of Principles of Building AI Agents um which if you're not familiar with is uh CEO Sam's book uh on basics of AI engineering um he actually just worked on a second edition of that book i think

it was released on Monday he's at the AI AI engineer world fair giving out copies so if you're there definitely go find him and uh get a copy but this is just another way to get a copy is reply to Shane's tweet with something else you'd like to see in the course and potentially win a copy yeah and uh get them while they're last they're uh flying off the shelf like hotcakes

yes yes oh and it's it's worth saying that you know it's also available for free online at mra.aibook I believe let me double check that yeah and we got we got Ward uh calling it a a Bible um or you know we can call it the Torah one or the other the book of M is what uh Abby says yeah cool that's what we'll go with um but

yeah okay so any any uh other AI news um Daniel and I just found out that Amazon is investing10 billion dollars in opening Yeah with the B data center 100 miles out of the Raleigh Durham Triangle area um how do we feel about that Daniel honestly I do not have big feelings one way or another about that um I just

heard about this like right before we jumped on so I haven't really been able to like take it in um but it kind of like I don't know not really surprising this is kind of I I think this is like what we're gonna see in the next many years uh just like a lot of money being thrown at like data centers a lot of money being thrown at AI kind of already seen that so this is just kind of the

the trend continued yeah definitely agreed um I know that Duke Energy which is the energy provider uh in North Carolina and in that area has been working with Richmond County North Carolina to improve the site i believe that they will probably be the ones powering the data center um I'm a big fan of uh Aloe Nuclear as well um I

believe they're researching ways to like actually coming up with new uh like micronuclear pods something like that i'm really not a physicist but um and they're targeting uh AI data centers as their target customer which I think is really smart um you know a lot of AI stuff is kind of speculative but I think

it's creating good economic pressure for innovation in energy that's Are you a lobbyist here are you trying to say something right now or I don't know i was just thinking I was like thinking you know like we're going to have so many more data centers um a lot of people get critical of the energy demands of AI so you know I think

it's nice it's actually like a real economic incentive to innovate in uh you know energy abundance yeah and I was actually talking to Tyler the other day about this also about just kind of like energy consumption in AI and it just kind of seems uh I don't know i mean like obviously biased but a little bit overblown just

because I think most new technology is very inefficient and so it's like right now AI is running very inefficiently and you're always seeing like uh like new models trying to run like more effic like the these companies don't want to spend as much as they have to spend on energy and it's like the more companies that are competing against each other they're going to have to figure out a way to run things uh more efficiently so

I think it's going to be like good in the in the long run yeah for sure um luckily at MRA I don't think we have to directly worry about that but uh you know what we do worry about is I don't know just making sure that people know how to get started using MRA so if you just joined the live stream um I'm Shria and this is Daniel we are uh I guess

engineers and marketers at Maestra um and we were just talking about the new course that we released uh I believe it's just master.ai/course um it's a course that you can take directly in your code editor definitely check it out uh I think it's one of the more like innovative ways to teach people nowadays

i'm trying to figure out how to uh pop some text up here but I don't know if I can i think it's captions captions we're also uh pretty new at this normally Shane or Obby does this second i think I I got captions i I got you oh yeah i got it i got it oh there we go yeah so if you just go there you can start the course check it

out um cool well I think we're actually just waiting on a few guests to show up uh I'm pretty excited about the lineup today we have Leo um who is a friend of mine who actually participated in the MRA Build Hackathon we have Rahul who is the founder of Julius um you may also know him from uh some press he got a few

years ago around when Twitter was having some layoffs uh I think we can go into that lore in a little bit when he joins and uh Tessa Barton who is an uh AI and machine learning engineer and she's also the founder of the GPU purse um you guys are probably wondering what the hell is a GPU purse but again stay tuned uh in a bit i have the same questions and I'm

psyched about it now would you get one now that you know what it is i'm I'm gonna ask for one i'm gonna put her on the spot yeah I'm gonna It's awesome yeah daniel did you actually go through the course yourself have I gone through the course no I haven't do you think like any new engineer who joins Ma this will be part of their onboarding uh that is a good idea

because like one of the things like when I first started like and I think this is kind of like what you do most places or like what you should be doing is that you want to get familiar with like the thing that you're building like you want to get like perspective of like being the user being like going through the steps finding all like the paper cuts

and things like that uh and it's always nice to to come at it with like a fresh set of eyes yes and then you also kind of like I don't know understand the whole story more like feel the the the customers or the users pain more things like that but yeah yeah for sure um I'm just checking in with Leo to see if he's on his way

hopefully he'll be here soon i think he is here oh okay you see him in the um Yeah I I see a man that says "Okay let me figure out how to do this." Is this Okay great there we go i was wondering because I was like "Yeah I'm sitting here." And you're like "Oh we're waiting and I'm sending him a message." And I'm like "Oh right here." feeling trapped uh trapped in the ether

so sorry about that Leo like we're new to this i know i know how that feels i just started doing some streams not too long ago and it is weird how complicated it is right like I didn't I didn't think about all that but uh yeah got to figure it out along the way right yeah um cool Leo okay so I I don't know what the best way to introduce you is i kind of know

that you are a jack of all trades so to speak yeah yeah i I do a lot of things um but Bestway is the founder of build and public university um so I'm really trying to figure out how we can focus education around building in public and kind of capturing some of these fastmoving trends and giving people like

the tools to um quickly be able to you know adapt to them and and use things like MRA to really go from like zero to one in building new products because like especially as technology evolves the easiest thing you can do is spend forever trying to figure out the right technology to instead of actually like focusing on

solving a problem um and so I really look at trying to figure out how to break things down into that smallest possible piece and yeah MRA's been great for that so uh awesome i was really excited uh when you joined Mastra because I was like "Oh that that that that's really cool because uh um love it." Yeah thanks go ahead Daniel oh did

did you build build in public university in public yeah yeah so it's actually been I've been getting it started for like the last uh three years now I think uh lot of starts and stops um but yeah just now kind of getting everything uh together and and figuring out um kind of the the last couple pieces with it and so uh

community's getting going now and actually we're having some fun with the uh Bolt hackathon that's uh that's going on this month and so actually I used MRA the MRA hackathon as a bit of a warm-up to kind of test some ideas and like share that with the community members to uh show them um how to do it and so like

I actually used uh used Mashra to replace like the the inner workings of like one of the community members products that she was building out and having trouble with so I was like hey here's the core functionality in MRA we're able to build it like nice and simply um and so I was able to use that as a demo of like Yeah can you talk a little bit Can you talk a little bit about what you built for the hackathon

and I think you've been working on some uh like so many different projects with Monster since and I believe you said there is one that you wanted to talk about today so tell us a little bit about what you did for the hackathon and then talk us through uh the project that you wanted to share today yeah so um

with the hackathon um I as I said I was really focused on kind of replacing the core workflow uh for my friend Jessica's product which was Indie Parrot and so the main workflow there is someone records a voice note uh it gets transcribed it um you know gets processed and basically turned into a newsletter so you can have someone go from like a voice memo to a newsletter

um without having to worry about that process and so for the hackathon um I basically just kind of broke down the steps into kind of you know the the transcription of it um kind of the the breaking out of the key topics um then I actually added in a step that can like uh scan a client's website and so like

if you don't have a scan it'll go get the markdown for the page um and bring that back in and then take kind of the core idea and then the the brand voice that got extracted from the website um and turn that into a newsletter and so you kind of have each of those steps um you know that that work together uh then in kind of that uh did you did you work

on that before or after uh we had the major workflows update right in the middle and so I was trying to figure out the new workflows um as as they were kind of uh being updated so the workflow part of it gave me a little trouble at the time but um the main flow uh was really yeah what mattered for it and so I didn't quite get like a nice demo of it for the hackathon but at least I got

the core functionality which let's face it when building products that's the part that matters right so I'd much rather have the functionality and not the cool demo than the other way around for sure um Oh we got a question from Ward oh uh are you joining the $1 million Bolt new hackathon um yes I am actually in

that one um and you can see uh some of the work I've been doing for that but uh my main entry for that is a time machine and I will uh talk about that um maybe here in a moment um yeah what was the what was the project that you did want to talk about today Leo well yeah so honestly it's um it's kind of related to the whole time machine um and really

time machine i'm like what are you talking about i can't like you know I I know and it's this has been the problem this conversation already I I I have had this conversation a lot um because this has been something I've been kind of working on for uh the past five years and uh what I realized is basically um

we don't do a good job of capturing the concept of time in computers right on the internet all time is flat and so what I've been looking at is how we can actually train AIs that actually like capture the element of time by collecting data alongside uh individuals that basically act as a growing context that you can use kind of alongside the

rapid evolution of AI models so what you do is you actually balance like you've got one side which is a growing context and kind of getting more and more stable over time as the uh models start developing faster and faster and faster and so like anytime you're building a system um as an engineer you don't want

a system that's changing all the time because that is exhausting to keep up with and anyone who's been an engineer in AI for longer than about a week at this point probably is like completely worn out because it's always changing um do you do you have like a like a demo or something to show us to kind of like um

like ground us visually or I feel like I'm I'm a very like visual person so I'd love to like be something even if it's like the code that you can walk through like I think that that could be fun so the the the hard part is actually that like I don't have necessarily the code for it because it's something I have to collect over time but I do have um an

experiment that I'm doing and so I started a new paid newsletter today um and as an experiment what I'm doing is basically I'm collecting uh microp payments to help out a friend of mine who uh is kind of struggling to pay some bills and so what this is is this actually acts as kind of a randomization um basically at like the launch of a

coin in crypto it you know you have everybody kind of buy at the same price but that doesn't actually capture the state of the economy at the time the coin was launched um and so what you want to do is at the beginning that a coin was launched you actually want to capture a bunch of randomized data that you can use to kind of anchor it to

reality at the time of launch um because then as it turns out you can do all sorts of really cool math stuff uh especially with AI um which is what I've been working on and I actually walked through it in the uh the first post of the newsletter um but yeah that you're using Maestra for um it it's I I'm

getting ready to to use Maestra to start um basically building out the systems that as the data starts coming in I can use to kind of process those um so as I learn it like one of the workflows I'm building with MRA actually is um a way to analyze someone's Twitter archive and get the entropy data from the Twitter

archive so like if you've posted a bunch of stuff on Twitter um you can actually use that to kind of see over time like how much growth you have in terms of like breadth and depth of topics and like relationships and so there's all of this like really cool detailed data you can end up deriving from kind of these like flat databases if you can start anchoring them um in different timelines

but you have to kind of like split those timelines out so what you do is basically you give individuals this growing context and then you can like it doesn't matter which AI system um it is kind of working with the context down the line you'll basically be able to use the same context for all models and so

it really like uh solves the whole like super rapid innovation thing because um as it turns out there's kind of like limits of how fast you can move because of the growing amount of complexity like as change increases you can't tell what's actually valuable and again in AI we kind of feel that every day so this should hopefully bring uh bring a good

bit of stability to kind of the AI ecosystem and let us all like start anchoring in some things that don't need to change so fast um so we can focus on creating real value as opposed to just spending all of that time keeping up with the latest tech as much fun as that can be because I know it is fun for nerds um and and I'm sure like a lot of

you I've been Yeah some some people like learning something new every week that's me i am addicted to learning and I will fully admit that uh you can do that so the rest of us don't have to is what you're saying well right that's what I that's what I've been trying to uh to focus on and really um and that's what the university is for right it's really

about like being able to kind of filter through the noise and kind of constantly be able to update the the knowledge base of like what's actually working what's not and so working together to really share that data as opposed to everybody kind of working on individual silos and like all building kind of the same thing over and over but like not very well because you know it's all moving so quickly so this

basically gives us like an open source way to really start building out entire ecosystems as opposed to just like individual projects so getting excited to uh to share some of the the stuff coming up with that soon awesome and it it looks like um we got a shout out from Obby what was your hackathon submission so the hackathon

submission was the Indie Parrot workflow um so that was the the one where I was uh yeah making the the newsletter from the transcript so um yeah that was uh as I said wasn't quite uh the the visual demo that I was hoping to maybe have by the end um but the the workflow worked and that was that was what mattered to

me so do you have that do you have a demo to that somewhere i don't know if we can um I I don't like I mean it's not a terribly exciting That's okay it's okay yeah it's okay yeah that's fine um Leo I'm going to I think I have your Twitter that I can blast out to people so they can follow along with your work um yeah awesome i have to admit I didn't

fully I think the whole you know time machine newsletter crypto thing it's a lot it's a lot of concepts in one thing um I I I I do have the the link to that if you want to see it and it outlines it a lot more clearly yeah I think maybe people can find you on Twitter if they want to follow okay uh ask some questions about it but Okay yeah perfect i I I will be there for joining us um

and I hope to see you in other future master hackathons that we do oh absolutely i love hackathons and I think it's a great time to uh to teach people how to to build cool products so um yeah uh we'll uh definitely stay in touch and uh let me know when the next one is cuz definitely it was nice to meet you too

bye Leo thoughts or reactions daniel I'm always like it's always just really interesting to hear what people are doing just cuz it's always so vastly different and when you're kind of like in the the headsp space of like building the framework you're not really like necessarily thinking about like all these different examples so it's like

when people come up with like crazy crazy yeah just like stuff that like like I'm not an entrepreneur i don't I don't think like like that so it's always like really refreshing that's actually a good uh good segue um we have Yeah I plan that ahead of time actually have this written down a a true entrepreneur joining the live stream it's actually it's nice to meet you i've only chatted with him a bit um on

Twitter and I think he I have to admit like many people he first came on my radar after his infamous performance art outside the Twitter headquarters um yeah so Rah how's it going welcome thank you for having me um and it's good to be here it's going well excited to chat with folks today are you at the um AI engineers world fair right now i am at

our Julius headquarters in San Francisco oh nice amazing yeah yeah um Daniel I actually told Daniel you gota you gotta like give some people some more context here you can't just drop it with uh like oh infamous prank and then walk away from that like these people want to know oh I I've told the story a bunch of times i will I'll let you guys tell the story yeah sure um I I think uh Rahul

and his friend uh Daniel were um standing right Daniel yeah yeah another Daniel just to be very clear different Daniel uh growing Daniel in fact um were uh at the right place at the right time i think when Twitter was having like big layoffs correct me if I'm wrong Rahul it's been it's been a few years and um I believe you were there with uh your Michelle Obama book and desk supplies

and you know the press uh were right there and they asked like you know who are you what was it like to get laid off um and I believe you went by a different name at that time uh it was a good experiment in how you can just easily manipulate the news I suppose and then Elon hired you briefly or something I

think i don't I don't I don't remember the exact details yeah elon didn't hire me but I did get to meet him a couple of times he's pretty cool and it was a good social engineering prank yeah yeah yeah okay anyway okay this is This is what This is what it was right that's you on the the left there yeah that's me yeah

that's I had long hair back then yeah um but okay uh let's let's talk about Julia so Rahul I know you're um you're from Singapore right and you immigrated here you participated in YC I think in 2022 you're the founder of Julius uh why don't you tell people a little bit about what Julius is julius is an AI data analyst you know people have a lot of data they have to

make datainformed decisions as a part of their job whether you work in startups you work in product management you work in marketing finance operations you have to make decisions with data every day and getting insights from that data is really hard because most people don't know how to write code they can't code

up an analysis and get insights on their own so we built Julius uh Julius is an AI data analyst that connects to your data you can ask questions to it get insights from it uh within seconds since launching in 2023 our users have made over 10 million data charts with Julius holy um every day Julius executes

over four million lines of analysis code and we're growing so do you want to show a demo a live demo are you Let me see if I have it pulled up um it's my first time using this this software yeah that's okay i understand if you have to like I know like sometimes Google Chrome asks you for permissions and then you have to um like

uh rejoin but I think Julius is so cool that in this case it's worth it if you rejoin yeah yeah cool you got to give the people what they came for oh it says share to screen oh it's working okay great cool uh let me uh so that's Julius and you know the easiest way to work with Julius is to just connect your

data and start asking questions so you know we're soon rolling out a Snowflake Postgress connector Bitcoin connector but for the sake of the demo I'm just going to show a a data set that a test data set it looks like transactions from a tool platform like Stripe so you have transaction data it's going to say load this data and show me a few

rows and the AI will take my instructions and the just like a human analyst it will write code to do the analysis so um it's uh looking at the the data thinking about it for a little bit and then uh turning our instruction into code and executing that code how long have you been uh working on this for um almost two years two years

yeah and I believe um I believe I read online that uh a lot of like researchers and scientists are the customers or like users of Julius is that still true has that changed yeah we have a pretty you know we that we one thing we've realized since launching is how ubiquitous need is for analysis and how ubiquitous uh data is people just have so much data

you know so scientists they want to analyze their you know experiment data there's um businesses that want to analyze their marketing and operations data and finance data there's product teams that want to understand the trends and how they're us are using product and so on so um all that so you know uh we have a pretty broad customer base um so

here you know I loaded my data as we can see there's like 35,000 rows of data a lot of data in there you know I can what I can do is I can just like either ask one of the questions that Julia suggests over time and then and then see what what I get and you know we just launched collaboration so I can actually invite my colleagues into Julius and

collaborate with them on Amazon real time and then soon we're launching uh connectors so you'll be able to connect to your Snowflake and Postgress databases and so on yeah super sick um so okay so I actually have a a almost like a business question for you um so some of in some of my discussions with

friends it's like you know all these like little AI apps agent apps agentic apps like it's hard like they're really fun when you use them um but it's kind of hard to make like a very userfacing you know like B2B SAS kind of uh product and there's kind of a question in my head of like will some of the first like is some of the first you know successful use case for this going to be like

internal tooling and I think I read somewhere online that like you actually are working with a lot of like big companies who just want to invite like everyone from their team and company and like be able to share data and you know those that analysis so you know like I don't know I'm wondering how how have you thought about that what are your

what are your observations am I totally wrong here absolutely well so you know a lot of our users are are exactly that persona you know companies that have data and they want their employees to be able to query that data in an easy way so here Gospel daily revenue over time as we can see there's some some trends

there's cumulative revenue over time and then like monthly revenue and so on there's the split between you know the different pricing plans and so on it's like a test data set but um what it shows is that within a couple minutes we were able to feed it the data and get these insights by just asking two

prompts and um that's how easy it is to do analysis with Julius and so businesses have a ton of data and internally they want to make it really easy for their employees and their teams and their workforce to get the insights that they need so for example you're a marketer you want to analyze how your

different marketing campaigns are performing for different segments of users sure like the data that lives in you know Google ads or Facebook ads you can pull that into Julius join that with the data in your database on like how the different users are using that using your product and then get insights

instantly by asking a few questions and because the medium of doing analysis is just English it makes it really really accessible anyone who can ask good questions can use Julius to get insights that's cool do do people ask for like uh like anything real time or is this mostly for uh like like static data that

you like fetch from somewhere or is there like a a a case for like building like real time dashboards with this as well great question so yes we get asked for that all the time um so we're building connectors so the first step in that direction is we're going to launch live connectors for data you can have

you know live data coming in from your snowflake and your databases and so on data warehouses and after that we're going to build dashboards so instead of a static dash like this chart you're going to have like a real-time chart you can just come back and consume over and over that's cool yeah um Daniel do you have any more questions from an engineering perspective

otherwise I can I have something else I I wanted to ask you know what go i can see you're you're really you're ready for it to do it um no uh I think Okay so you know Rah I actually I write I was reading your profile um in Arena i actually write for them sometimes as well um really love that profile um and I think there's an excerpt in that article about how uh for you like the

entry to Silicon Valley and like networking and getting jobs and connections was through hackathons um and I believe you went to like UT Dallas i think that Yeah okay yeah um and I feel like there's like this like is there you know are we having a hackathon comeback and renaissance like what what's going on what have you observed

just just share your thoughts yeah yeah i really hope there is a hackathon renaissance you know just the energy of being in a room for 24 to 48 hours where people are just trying to make their idea a reality it's really special and that's what really got me into startups and building back in college you know you you have your you have your your lectures you have your

assignments and that's really boring and really mundane but how about you take an idea and take 48 hours to make that idea a reality and you like run into bottlenecks and and roadblocks and you sort of like find ways around it to just make it work um have you like hosted any hackathons before or just like

participate sorry i'm I'm a I always attend yeah I really like attending hackathons i'm always gonna have one though we should we should do a data thon yeah we should do a data th where you can just analyze you know get really really cool insights especially with like you know with the tariffs or like

macro trends and global things that are happening around the world like can you just find interesting data and like get insights within within minutes yeah is there Sorry Daniel I have I have a quick question like do you uh is there an interesting question that you had for a long time that you were able to answer

with Julius like you know I don't know what's like a bit like an aha moment that you personally have had yeah well you know we use it all the time internally you know for for our growth stuff um you know we we do a lot of SEO so like if you look up AI data analysis Julius is number one on Google in fact if you look up Julius we rank above

Julius Caesar um and so for all that like we we use Julius um to you know do a lot of our growth analytics understand deeply how um you know the data what was the market data look like what is the growth you know SEO trends look like and so um that's we use those pretty pretty often and so we wanted to build something that we want to use and we find useful yeah yeah i

guess I guess you mentioned like you know like like economic data tariffs etc i was wondering if there's like a fun question like that that you know that you've like just spent like a few minutes going down the rabbit hole with Julius exploring yeah I I do all the time um there's there's a there's like a really cool um

visualization that one of our users shared where he he created a time series animation of like Roger Federer uh Jookovic and um was the third one Nadalionships over time and like how they were competing like you know the three great uh I thought that was pretty cool another user recently shared a um chart of like secretary of defense tenur

years like shortest seg 10 years I believe that's pretty cool so there's always like interesting things that people like are able to do with data especially with like current current trends current affairs always fascinating probably useful for making some viral tweets or something i don't know yeah should get back on Twitter i haven't been on Twitter i think I Yeah I think I tweeted about this live stream and I was like you know

Rahul performance artist or something and I tap into that side as well totally absolutely sorry i wanted to go ahead yeah i wanted to ask something uh when we were talking about hackathons um as like somebody who attends a lot of them do you have any like tips for people because I think like like we we've done our own hackathon we're we're going to do another one at some point i

think a lot of like our community and audience are the types that really enjoy going to hackathons do you have any like suggestions on like how to get the most out of going to a hackathon or like what do you what do you like look for like how do you make it like enjoyable for yourself and does it differ for like

online versus in person i yeah honestly I haven't done a lot of online hackathons so I don't have much tips there but I think the best part of hackathons is like the people you meet is is I think the part of hackathon is have fun meet people that are working on fun things um our CTO Matt actually met

him at a hackathon three two and a half years ago and then I spent three months recruiting him after that what was the hackathon i think it was the scale AI hackathon January 2023 and what made you guys click like did you work together on a project or Yeah no we did not but I was just at the hackathon and I saw his project and I was like this is so cool like uh you know and I just like hung

out with him all afternoon just like trying to learn more about his project and I was like this is this is so cool like I'm having a lot of fun talking to this person like I really want to work with this person um I think a lot of you know recruiting at startups is like do you walk when you talk to somebody do you walk away with more energy than you like went to the conversation with and

then and then um you just like you know working on cool stuff just for the sake of fun and um I think that's important about startups is like you're building a project that's not like super serious like oh here's a payment processor that helps you detect fraud uh or whatever it's it's more like you know the first I

remember the first hackathon I went to this this person he made a um he made a bot that would play the Chrome like T-Rex game like this on your screen and then like trigger a keyboard uh click every time like you see like a object come up and I thought that was pretty cool like it has like no purpose is so cool that you can do that um so I think

the point of hackathon is to have fun and meet cool people yeah i always think I like that i always think that companies should do um internal hackathons as well like you know take a few days just let everyone in the company whatever you know we always have ideas when working like of like oh I wish I had time to you know tinker around and use Monster to build this internal tooling or something else but

like isn't the famous example that like Google Maps was created during a hackathon within Google wow I didn't know that that's cool i'm like I'm like 95% sure I just didn't make that up right now someone fact check this yeah where's our live fact checker yeah but anyway thank you so much for joining Rahul hopefully I I can imagine that um

we'll probably ask you to join some future live streams and you know whenever whenever you want you're welcome back yeah great thank you so much for having me it was great meeting you yeah nice meeting you cool take care bye all that was really fun honestly I've been like dying to meet him for so long um

yeah that this uh this stream is just for you to to meet all the people that you've had on your list of like people that you want to meet exactly next is like Michelle Obama then um no next is uh Tessa Barton people were expecting Michelle Obama you're going to be disappointed no no it's it's all good i think you're doing something super super cool um yeah you

know I don't think Michelle Obama made purses so uh yeah I want to fact check this stream maybe we were wrong but um anyway yeah um okay so we have Tessa um Tessa I've met I think it was at a conference like at Manifest or something yeah manifest yeah where I think you were actually selling these purses um so

I also want to clarify that you are also an AI machine learning researcher and engineer so uh I'd like to you know learn a bit about that but Pier tell us about these purses that we've been dying to know about okay so starting like two years ago I got this really strong compulsion to make a purse with a GPU on it because I had this problem yeah i'm

going to pull this up because these are pretty amazing yeah oh my gosh those photos so I just was thinking about GPUs all the time and also I'm like a total shopaholic i love purses um and the it just came to me and so I started making them um I bought a bunch of like ingredients on Amazon and uh after like carrying them around for like a summer

and selling them at Manifest they just blew up at New York Fashion Week last year um and I was really not expecting it um and in fact I didn't even have really inventory i had like huge delays on shipping for the first about the the New York Fashion Week story i think I I missed that i think I I saw you at Manifest and then I saw some news

articles and was like "Whoa." I was trying to sell them at Manifest i was begging people to take them just like it was totally missed i remember I remember you were like for a nice lady like you I'll cut you a deal like 300 no I was literally losing money on the purses for like for literally months just cuz the GPUs are expensive so like I was uh just

like giving them away selling them giving away as presents and I actually gave one to my former coworker Christina and she posted it and took it around New York Fashion Week and it just totally blew up on Twitter got like millions of likes um she took like a kind of sexy photo with it and inadvertently like

launched a really awesome trajectory and so like I was scrambling to like put together a Shopify storefront in those hours after the tweet went viral i think the way you were selling it at manifest that I can recall is like approaching approaching men and being like you want your girlfriend to be walking around San Francisco with this Nvidia purse and

like that's you know Yeah you don't want her to be the only one without one exactly oh my gosh yeah so I really wanted it to be like the Cuz like Right if you show up to a San Francisco party with a Chanel bag very few people are going to appreciate it and the ones that do it's just not the right audience if you show up with a nice GPU that has like flashing lights it's just get so

many more um how are you how did you source the GPUs so the GPUs so I have this friend who has like a bunch of lasers in his backyard and that's where I got the initial chips because he just had a bunch that he was like disassembling and then the other ones after that I just get them off eBay or wherever I can get one so there are super some people have like GPUs that

they are really into like um that have nostalgic value and so I've been trying to like source those um I got a pink one from Japan that's selling well so is every purse kind of unique like each of these GPUs are are different yeah they're pretty unique like some of them I have to cut down to size it's not like

for better for worse it's not like a very standardized process but Yeah so they're like one of a kind essentially like two people with the same purse it's going to look different yeah well in fact it's actually good that they all look different because I had a batch of GPU purses stolen from my car so if you see anyone with the black Nvidia GPU

purse that's a stolen merchandise right there damn damn um Tessa tell us a little bit about your career as well um so I have been in AI for a while i started out at Snapchat working on face filters actually um in 2017 and one of the face filters I actually made the first GANbased filter that went to a million views and it was used to catch a

pedophile so like that was like a kind of weird career point so what what can't you do no I mean like I then went Yeah i mean purses catching pedophiles um like uh then I' I've been at as well like New York Times sports analysis yeah okay totally random i started doing like sports analysis of photos using computer vision um we discovered a bunch of

really interesting stuff that I think most people don't know about like sports because most sports Yeah if you apply AI to sports is kind of crazy like we discovered that swimmers even at the top level they slow down throughout the race if you look at the frame by frame they are getting slower each lap and each lap as they progress is slower so like I

think is that the same with runners is that the same no it's not the same with runners runners speed up at the end almost every other sport if you look at like fine grain analysis of the speed using uh video and vision models like they speed up but swimmers swimmers just like slow down the entire time yeah

swimming is about not losing time it's not about speeding up so I I I thought that was very interesting so I'm really not losing time versus speeding up that's what I kind of think after doing this analysis interesting and we actually shared this with like the Olympic athletes coaches and like they didn't know because like there's a lot of stuff that you just can't analyze unless you have really fine grain analysis but that's a whole other like

Yeah no I want to hear more about that do you have any other uh any fun any Yeah oh my gosh like at the Tokyo Olympics we were so screwed because our vision based system got so confused because it rained and so you could see like the reflection of the runners off the track that was happened um other than that I have been working in B2B SAS at data bricks as a research scientist and that's much more standard

and um it was it's so much fun yeah yeah do you have any uh hot takes on AI oh I I Anything about agents anything about agents thank you thank you yeah just go go through your list uh give us every hot take you have and uh we'll we'll be here for a while i don't know if I prepared any hot takes oh that's okay it's okay oh my gosh any cold takes

lukewarm we'll take them all you'll take them all okay agents like I really just want an agent this is so bad but I never mind i shouldn't share this idea i can't i can't someone needs to build an agent that like goes through and finds all the haircut appointments in an area and like books you one cuz right now that doesn't exist

so it's like very hard to get a haircut no it's just like you have to go click through to the end of the booking process to find out if there's availability wait so what is that haircut site that people use vag Vagaro or something do you know what I'm talking no centralized as far as I know site for searching for

a haircut like for me people not have like a trusted hairdresser where they're like I only go to to my person i'm not going to like shop around and go wherever it's like I have like Jasmine she's the only one that cuts my hair she's the only one I trust yeah definitely definitely um but sometimes you really just need a haircut like

sometimes yeah I think need I think we need FCP for uh I use I've used Vera before to find like it's almost like um I don't know if you know MindBody it's like the yoga studio booking app uh it connects to like a bunch of different yoga studios and you can just like find something like today at XM where what

are the classes near me uh I think Vaggera is a similar thing but I guess we need MCP and then we need to give it to an agent and have it book our haircut appointments yeah seriously maybe I should just work on that today instead of Yeah you should i mean maybe use try use mostro work that would be trick

it's honestly insane how fast stuff is coming out though like But it's also kind of insane what hasn't been built yet so I think that's interesting place to be yeah we barely we barely started i mean I didn't know that GPU versus like like once it was a thing I was like "Oh so obviously like this makes so much sense." But you were the first person to

really to get there and do it well I mean I invented the concept i don't know like other people I I don't know uh there are copycats it's a little bit like awkward but whatever it's fine we We can call them out here we can we can shame some people it's fine it's fine um thank you so much for inviting me on this this is really fun no thanks so

much for coming on Tessa um I'll never look at Olympic swimming again the same now like it's Yeah I'm going to be watching Extra Close now exactly exactly yeah i mean I actually think the next Olympics will be really interesting because NBC is finally like catching up with like computer vision and they're

like doing mocap so it will be an interesting next Olympics oh and uh you will not be disappointed this is where people can find you on X i think I I got that anyway thanks so much for joining Tess thank you have a nice day thanks bye bye any reactions Daniel how do you feel about today's uh today's lineup of guests it goes so fast yeah i honestly

cuz we were talking about like Oh cuz sometimes it's like two hours and we're like okay we're going to this is our first time we're going to do like like an hour hour and a bit maybe but honestly it flew by and I'm like you know what i understand how Shane can do two hours two hours i agree i think but if it's like by himself like oh my god I

would not I would not do two hours by myself absolutely yeah um I mean the I don't know the sports analysis thing is actually really interesting it's not it's not intuitive to me that it would be so hard to have those insights but apparently like I guess it's image image capture image analysis that's difficult maybe Tessa will build the Julius of image analysis

or something i could see that happening yeah she she needs a blog that I can uh look up all of her sports sports facts oh true true for sure yeah cuz I cuz I'm basically I'm going to repeat this at like a like a party or something like that to impress people so I need like a good repertoire of like like random

sports facts exactly exactly that's how to make sports interesting is with AI I suppose how to make Yeah um okay closing remarks uh you know I don't I don't have a lot to think about um I think we mentioned at the beginning of the stream that we released the new master course uh Daniel you have your little Oh I think I found it i found the little caption um it is I

think the first course uh that you can take right in your code editor super super cool so it's powered by our MCP doc server um so if you go there you'll you'll it'll tell you how to get set up in cursor wind surf whatever your agent of choice is yes um and I think we were going to do a giveaway uh giving away uh copies of the

newest edition of principle of built principles of building AI agents i'm actually going to I should probably and I know you're probably thinking didn't the first edition just come out in a few like a few months ago and the answer is yes but in the AI world a few months is years so there was enough to update it and I

think what do they say it was like 50% more information or something like that something like that yeah um but yeah um so we're giving away print copies of those uh just respond to Shane's tweet storm here it is uh I believe the last tweet is asking for uh ideas and feedback you know like what lessons should we add next so we're

going to be going through any replies and picking out a few people to reach out to and mail them a copy of the book um of course you can always find uh an online digital copy at master.ai/book as well um any other parting thoughts Daniel should we just wrap it up yeah yeah we'll we'll wrap it up yeah

yeah okay well thanks so much for joining um and hopefully we'll see you at some other point in time who knows when yeah death