SPEAKER_00: oh jason where do we even begin your tweets are like a roller coaster ride through the twitter verse one moment you're making predictions about banks pretty good paul and the next you're trying to convince us that a tiny covert pendant isn't the creepiest thing ever okay that's pretty good those are good polls yeah let's not forget your obsession with cold plunges you're like a human popsicle oh always trying to one up your friends on twitter and don't even get me started on your ai request you're like a kid in a candy store except a candy store is twitter and twitter is an ai generated renaissance painting but hey let's uh at least you're consistent in your love of roasting people just remember that jason you're a prime example of why we should never take ourselves too seriously on the twitter keep the laughs coming and maybe one day you'll find a cold plunge that can SPEAKER_07: watch away the absurdity of it all this week in startups is brought to you by linkedin marketing to redeem a free 100 linkedin ad credit and launch your first campaign go to linkedin.com SPEAKER_08: slash this week in startups vanta compliance and security shouldn't be a deal breaker for startups to win new business vanta makes it easy for companies to get a sock to report fast twist listeners can get one thousand dollars off for a limited time at vanta.com twist and coriant real wealth requires real solutions coriant provides wealth management services centered around you for more information speak with an advisor today at coriant.com all right everybody welcome to this SPEAKER_11: week in startups it's monday so it's madra monday yes my pal sun deep sunny madra is with us he is SPEAKER_13: from definitive intelligence and i'm an investor in that company go check out definitive.io and what we do here every monday now in the age of ai you can get used to it is we just show and tell demo or die like the old days in the uh computer hacking clubs here in the valley you would just show SPEAKER_15: what you built you'd show your work right you'd show up you show up with your work and you share your SPEAKER_13: work and uh we all give it a a letter grade here and we give it some feedback and some general thoughts so this has been a crazy week you and i have been for the last 10 days playing with grok g-r-o-k the new uh ai startup by elon musk and uh a really impressive group of talented people and uh they have access to the twitter fire hose which nobody else has access to and so let's get right into it sunny what are your initial thoughts and maybe we could do some demos here at grok SPEAKER_22: what's working what's not yeah it's very early days so frame it for us where they're at yeah SPEAKER_23: well very early days like first of all there's a couple of things that you know we got to give them SPEAKER_24: kudos for like you know the company was only ideated you know six months ago even less like i think may and then uh or maybe that's six now and then honestly i think they really only started building it like three months ago and so this is version zero that we have access to and i think they're SPEAKER_26: already working on version one got it um they've talked about the infrastructure which i think elon SPEAKER_24: mentioned on the lex pod the number 400 which they did this with like 10 000 a 100s right so yeah um so you know really interesting and that is a large cluster of a ones it is it's you know it's SPEAKER_26: probably you know like 100 plus million dollars of capital expense or if you wanted to you know rent it from someone probably cost you similar amounts on someone's cloud and hard to get access to that so yeah he probably knows a few people given who he is to get access to it and build that's SPEAKER_33: that and i suppose if elon falls and wants some a100s maybe he uh can get the right person on the SPEAKER_26: phone yeah and look like they have the advantage that they're they're not new to this right they're obviously been building clusters for a long time uh within tesla right and so yes they there's a there's some advantage to them that they understand like how to build it how to get the internet connects SPEAKER_40: all set up and you know data center space and everything else so it just gives them the advantage SPEAKER_44: of moving pretty quickly because sometimes you can get stalled there right just infrastructure once again is a roadblock it became a non-issue for the last 20 years since cloud computing came out every month it gets cheaper and easier to just throw something up on the on the web on the internet now with ai a little more compute intensive not so easy i'm sure we will build up the resources in the next i don't know two years three years for it to be easy again do you think or do you think we're SPEAKER_26: going to constantly be playing catch up here well you know this is what um elon i'll just kind of go up because we're going to get into grok here but he mentioned on the lex pod that he thinks we're going to have an oversupply of ai related compute resources within the next 18 months that was my SPEAKER_13: position as well uh yeah and i talked to him about it and uh i use the analogy of fiber remember we got all got very excited about fiber in the uh end of the millennium uh the end of the center last century and they just started building so much fiber everywhere and we didn't exactly have the tools uh you know built in we didn't have youtube we didn't have hd cameras on our phones right so what was the use case for all that bandwidth and a bunch of companies went out of business and then a bunch of that got bought out so they have it and i'm just pulling up this tweet just to get it right so i SPEAKER_24: think stars eight eight thousand not ten thousand eight hundred and that he expects uh you know so SPEAKER_59: he expects them to get to right and so 64 000 by mid next year pretty 64 800 by next year wow okay so SPEAKER_25: here we go everybody if you're not watching we're going to sportscast this so we'll describe what's on the screen uh for you and the implications of it but if you want to watch uh spotify supports this SPEAKER_44: week in startups video there is a this week in startups video feed on itunes if you search for SPEAKER_63: it and of course youtube just search for this week in startups okay here we go so you know one of the SPEAKER_26: things that we'll just start out with is that you know it has these like two modes and so if i if you know if i kind of start a new chat and i'll flip back and forth here a bit you're going to see fun mode and regular mode got it and and you know the fun mode as you'd expect it kind of takes things takes an edge off makes them kind of funny and you know one thing that we must do i think um and i've got some you SPEAKER_24: know queued up already so you know we don't have to wait for a different it's it's quite fast for some SPEAKER_29: real time but i said look one thing that everyone's doing last week jacal no one did it and now we'll do it this way is like write a toast uh write write a roast of twitter user at jason based on his tweets SPEAKER_53: love it jason the best thing to do is for you read your own roast here so i'm gonna let you do it SPEAKER_00: oh jason where do we even begin your tweets are like a roller coaster ride through the twitterverse one moment you're making predictions about banks pretty good paul and the next you're trying to convince us that a tiny covert pendant isn't the creepiest thing ever okay that's pretty good those are good polls yeah let's not forget your obsession with cold plunges you're like a human popsicle oh always trying to one-up your friends on twitter and don't even get me started on your ai request you're like a kid in a candy store except a candy store is twitter and twitter is an ai generated renaissance painting but hey let's uh at least you're consistent in your love of roasting people just remember that jason you're a prime example of why we should never take ourselves too seriously on the twitter keep the laughs coming and maybe one day you'll find a cold plunge that can wash away the absurdity of it all i mean listen it's not ready for stage yeah but it's a good workshop like if you SPEAKER_13: had a junior comedian in a room and they said hey we should do something about cold plunge and that the ai pendant that you were wailing on and of course your silicon valley bank all caps tweet that everybody roasted pretty good start honestly and uh you can see the power of the twitter data and the more involved in social media you are what you can see is ai is going to you know with access to the latest and greatest on social media which twitter has a lot of the smart people this could be like this SPEAKER_44: could vindicate elon's purchase of twitter now he purchased it because he wanted to get rid of the SPEAKER_73: woke mind virus and because i think he uh you know had a personal affinity towards it he loved it but he did think that you know uh freedom of speech was important this woke mind virus needed to be SPEAKER_75: killed and i think also now you can see that hey maybe he won't lose 20 billion dollars on this maybe SPEAKER_44: this is the start of something really important and if he has this data and nobody else is allowed to SPEAKER_78: have this does then you're going to be using grok an awful lot right you can see yourself saying go use SPEAKER_26: grok right well that and you know especially and like let's just maybe kind of keep going through these SPEAKER_24: examples where we can kind of highlight so you know last week there was um a controversy around the latest mr beast video where he um you know made a bunch of wells in in africa yeah and uh so you SPEAKER_81: people clean drinking water and the mids on twitter attacked him i think honestly i don't think these SPEAKER_13: people who are attacking him actually think this i think it's just an engagement farming strategy totally totally it's just the way like who could be upset about giving water wells to people in various SPEAKER_29: african countries yeah yeah yeah and what's really interesting here is like i like i like the approach SPEAKER_26: that crock took this is obviously not in fun mode this is in regular mode where it kind of gave both sides right it told it said what the critics said and then it on the other hand it talked about SPEAKER_10: you know the what the supporters of mr b said so you said what is the controversy with the latest mr beast SPEAKER_13: video so yeah this is something where you will see something trending and you want to work backwards to know what's going on yeah you hear somebody was on trial somebody died yeah a celebrity died somebody's trending and you're like oh my god christopher walk is trending oh no christopher walk is trending SPEAKER_98: wow yeah and you get nervous right and you want to work backwards so here you're just asking grok SPEAKER_44: what what is it and it says the country surrounding mr beast latest video involves his philanthropic efforts in africa he recently built 100 water wells in various african countries which SPEAKER_38: has sparked debate about the portrayal of africa and the impact of such actions on local communities SPEAKER_13: oh my god when people when people say local communities and uh perpetuating stereotypes that's when you know you're going to get snowed right like there's not actually an issue here there's no SPEAKER_44: stereotype being perpetrated here except it's the frontier market it's an emerging market and emerging markets are further behind the developed markets and that can go for china vietnam and SPEAKER_75: then to an african country then to you know you know a country run by a dictator like north korea SPEAKER_44: there's no harm being done here to say that some places need clean water or maybe they don't have SPEAKER_106: access to clean water oh my god critics argue that his actions may perpetuate the stereotype that africa is dependent on handouts and quotes and could potentially shame the kenyan government well if the government's SPEAKER_109: not giving people clean water then maybe they should be shamed yeah i mean i hate to i hate to point SPEAKER_75: out the obvious i hate to be peckin obvious on the pod yeah yeah they believe that this approach is SPEAKER_44: charity this approach to charity may not address the root causes of the issues faced by these communities okay you know what the root cause is these countries are further behind in their development because in a lot of cases they're run by dictators who have stolen all the money and have oppressed their own people in which case that maybe you do want to shame them now i'm not speaking about kenya SPEAKER_73: in specific here but i'm just talking generally about dictatorships anywhere on the globe it could be latin america it could be africa it could be asia it could be north america on the other hand SPEAKER_63: supporters to mr b appreciate his efforts to provide clean drinking water to 500 000 people and believe that SPEAKER_117: his actions are genuinely helping those in need and then it gives you two tweets that this is called SPEAKER_44: from and you've got to assume that it's pulling not only the tweet but do you think it's going to the SPEAKER_23: target news site and pulling that as well uh i think like like you know as we go through more of the SPEAKER_29: examples it has the capability to do that like when you when you and you know we we can do some live SPEAKER_26: i just pulled these up so we don't have to kind of do do them as we do but like i have some fun ones here that kind of highlight it as well so okay this is another place where like you know twitter SPEAKER_24: uh sorry grok is really shining because of its access to twitter data um and and not having to i guess you know rely on like web search which is going to lead to um you know you know i think problems in in some regard as well with like all this you know the people there's a new version of robots.txt that's out there right where sites are saying hey don't crawl my my data for ai SPEAKER_59: bots right that was kind of implemented recently and so was it what's it called you know does it have SPEAKER_29: like a name well i don't want to kind of misstate it so we'll pull it pull up the actuals but it's SPEAKER_50: like you know like robots.txt but for for the eyes i said that a year ago like they should you should SPEAKER_13: have to there should be a license so it should be you know whatever it is newyorktimes.com license and when you read the license it should say in the license in html readable semantic code yeah here's the license you can use it uh you can reference the new york times in your language model with a citation and you can summarize an article of up to um you know for up to two sentences of summary three sentences of summary you cannot use it to train your model yeah right so training equals off SPEAKER_44: exactly yeah or it could be you know for every if you want to you know summarize this you have to SPEAKER_140: have a paid subscription yeah any number of issues any number of ways to handle i was just pulling SPEAKER_24: it up here so i think they're just doing it within uh robots.txt you can block the gpt bot user agent so they've given it a different user agent that you can put in there and say hey you know you're not SPEAKER_38: you're allowed to come here i did that proactively yeah yeah oh smart move on their part so here comes their defense hey you can just opt out it's up to you so that's the that's the opt out of a defense SPEAKER_75: it doesn't work um it's just on a go forward basis it might lower the number of people who might SPEAKER_149: potentially sue you doesn't mean if you trained it on data you didn't own that you're not in trouble SPEAKER_150: yeah all right business to business founders need to understand two very important things b2b marketing is hard we all know that it's really hard number two it's simple on linkedin ads SPEAKER_153: why is b2b marketing so hard because the decision makers hey people like me right the founder of the company or maybe you you're an executive listening to the show they're hard to identify right and they're really hard to target over the internet and how do you know you're reaching them in the most respectful environment people who are decision makers you cannot reach them through that 1990s way of reaching executives no they need social outreach and the way to get that social outreach dialed in is with linkedin linkedin solved this problem there are 70 million decision makers hanging out on linkedin and executives like me yeah the decision makers we're on linkedin we're deep into linkedin every single day and linkedin ads helps you get your b2b message in front of the right people like me and listen when people are on linkedin they want to do business linkedin equals business business equals linkedin each social network has its vibe and the vibes at linkedin are business it's business vibes only and it's happened finally linkedin has hit one billion users the march to a billion has been achieved congratulations to all my friends at linkedin so here's a call to action get started today see why linkedin is the place to be for b2b we'll even give you a hundy right now a hundred dollar credit on your next campaign go to linkedin.com slash this week in startups to claim your hundred dollar credit why wouldn't you take the hundred go to linkedin.com slash this week in startups the links are in the show notes terms and conditions do apply you know here i'll give SPEAKER_36: another like a good sports example where you kind of want something much more real time and so remember SPEAKER_26: some people are listening so read it yeah exactly and i'm going to go through it so i basically said hey how did zach wilson play last night you know he's been the replacement for aaron rogers and he says you know based on his information available zach wilson's performance last night was a mix of positive and negative first player to have 250 pass yards and 50 rush yards which is notable but he also threw an interception late in the game which was considered inexcusable by some SPEAKER_29: and so you know really um just super helpful like for these type of things where you know something has happened you're seeing about it and then it gives you the citations it shows you the tweets SPEAKER_162: it pulled this from yeah exactly see that's the key this is what i've always wanted yeah i've always SPEAKER_13: wanted the ability to summarize what's on twitter in real time it's really powerful you know like tell me the sentiment of the movie the killer which we were talking about in our group chat yeah you know some people liked it some people didn't who liked it who didn't and then you could would be really nice is if i could say here which of my followers you know who are the followers of that chase and who like the knicks i would like to know who follows me that likes the knicks can it do that i don't think it can do that yet it doesn't understand followers follower counts yeah you know the database of tweets like uh what would you call that information like sql information database kind of information SPEAKER_24: the the metadata of tweets let's call it that right you know because that's the you know who liked it who's followed it right and metadata of users they haven't kind of made that information available to grok yet which is um there and then look like you know there are some other things i'll just do one or SPEAKER_29: two more and we'll go on from here but like um you know this is the one that's been moving around SPEAKER_26: and i and i think it's been let's we can maybe come back to this one next week where i know you were at the first one i really would like to go to this starship launch because you mentioned it was like SPEAKER_75: one of the most incredible things you've ever felt from the uh it was incredibly touching i guess a SPEAKER_44: human to see that rocket ship go up and think about the ramifications of being able to send hundreds of humans cargo whatever to mars it was like very touching in a weird hard way to describe but hope SPEAKER_177: would be the overwhelming hope and joy for the people who created it yeah okay yeah and and just SPEAKER_24: uh i heard like you know from folks that were there including yourself that it was just the sheer force of sound and energy was even you know you guys were several miles away or you know and uh you could still feel it kind of through in your boats all right and so like you know this is one where i think it's SPEAKER_26: doing a pretty good job um and then you know there are some cases where you know we'll just go to kind SPEAKER_29: of last week's pod which was the one the all in pod which is the one with with jared kushner and in this case i said hey can you provide a summary that's some interesting feedback um can you provide a summary of this week's all in podcast that sure you know can you tell me a specific episode so that was episode 153 we did ask you for that huh yeah it did but you can see here SPEAKER_106: it got it way wrong right and um yeah it's impact on job svb collapse that was yeah that's super yeah SPEAKER_26: and you can see here the hallucination is strong because it's like being very kind of affirmative in like it thinks that episode was about that it had you know svp and kind of mixed a bunch of things up i don't even think this this episode exists and obviously in this format right i think that this SPEAKER_13: is the balkanization in some ways of what we're going to see in ai which is this would be a much better uh query for either um somebody made an all-in ai database which is just the episodes uh you might have seen podcast ai a company we invested in um and then other people have started to make all in transcript thing you know ai versions where you can just only search the archive it would those would nail it right and then you have youtube which has it up there but then you start thinking about spotify and you start thinking about itunes if spotify and itunes took the transcripts everything SPEAKER_73: i don't know if they have the legal ability to do this i think you would have to ask us for permission to build a derivative product but maybe they'll just build it anyway um you know you could be SPEAKER_26: searching for that yeah yeah we'll close out on a fun one and so i just dropped in 420 this number means a lot of things a lot of different people yeah and it gave a really great response here ah the number 420 a true symbol of the universe's infinite wisdom in some cultures it represents the ultimate answer to the question of life the universe and everything in other cultures it represents a time of day when people gather to discuss the mysteries of the cosmos yeah they do that too sort of yeah SPEAKER_196: yeah i mean you know it's an interesting time it's almost quitting time yeah at 420 you could do SPEAKER_177: a lot of different things we yeah we had kids it's a family show here so let's keep it yeah i SPEAKER_29: i really like uh you know like i said i really like its ability to have access to data outside SPEAKER_26: um of just its kind of training corpus and the way it answers questions so uh i'm gonna let you lead out jacal what's your grade on this one um okay so it's not perfect so SPEAKER_98: you can't give it an a no yeah but it's certainly not bad so you can't give it a c Chamath Palihapitiya: which means it's going to be a b right okay so now you're gonna we argue b minus b plus or b SPEAKER_13: right that's i think any logical person who's listened to us do this this and this is a 0.1 product so based on how much work it's been into you know has been put into it you could qualify it with that and say for three or four months work this is an a plus you could say that but i'm not grading it based on how much works when i'm grading it against the market so grading against the market i wouldn't give it a b plus yet but i wouldn't give it a b minus so i'm just i think it's a solid d like out of the gate it feels like you can get use out of it but out of the gate it's not gonna have the complete set of twitter data that i want yeah and i want to be able to say uh tell me who is the most who are the most credible twitter users uh talking about covid or SPEAKER_44: who thought the vaccines were overhyped before you know we had the data for it right like something really sophisticated like that which takes into account time the user account how often they tweet about it how often they get retweeted how much engagement so if i could just say hey who are SPEAKER_75: the most engaging um twitter handles um for the next that would be really interesting for me um and SPEAKER_44: if it actually gave and then put that on a twitter list and follow those you know like that's what i want to do i kind of want to automate what i do on twitter um and there was a company we had invested in called little bird it was get little bird and they got bought by another company's uh sprinkler uh and these companies did things like this with the twitter api not using ai but they would just look for keywords and then they would build an index and they could tell you these are the most followed accounts talking about the next these are the ones that get the most engagement these are the most engaging repliers you know replying to them so you could kind of figure out and then you could say if these were the top 100 nix accounts who follows the most who who of the hundred gets followed the most by the other 99 so then you can build like a meta ranking so these are the ones talking getting the most engagement overall talking about the nicks but if you take those hundred who follows each other you're like oh and this person is the the goat of the goats you know it kind of could be very SPEAKER_215: interesting yeah um yeah i also like being able to do multiple searches at once i thought that was SPEAKER_24: kind of cool yeah like you know you i'm just kind of showing it here for those listening side by side running side by side and and you can even branch conversations right so you can look at the response tree of you know how it got there it's got some pretty interesting features you can regenerate the conversation um you know uh like i i tried what you were saying here and you know it does find kind SPEAKER_29: of the list of things but then it gives you like opening these are good right and say like hey uh SPEAKER_26: you know here's the way for you to kind of create a twitter list which is great but you want that all SPEAKER_29: integrated um look i'm going to take into account how quickly they've done it access to proprietary data and i'm going to put it at b plus i think okay fine i think to to do it that fast the speed at which it runs at and integrated kind of with the twitter experience and i think there's just some nice to halves but like you know as a first release and you know we're getting spoiled because we see the best of the best all the time right and so um they've done an incredible job kudos to the team they've done it really quickly they're taking feedback into account and there is a grok subscription plan part of the SPEAKER_13: subscription service costing 16 per month under the x premium plus so on x i think you get premium plus yeah which i think i just have premium yeah but i had a friend at twitter who got me into the grok beta but i think i have to go cancel my that's what i hate about subscriptions no you know i i moved SPEAKER_26: no but i am on itunes yeah subscribe even from itunes you can do it there you can upgrade within SPEAKER_29: the app you can upgrade from uh premium to premium plus and it does okay that's what i need to do okay SPEAKER_231: and it does like i want to go direct so they don't because don't they charge you the 30 extra SPEAKER_26: like i think twitter i don't know if they do 30 but they yeah there's a premium you pay for going SPEAKER_13: through the app store but yeah yeah i guess so if you that's an interesting just user question like do you want to go to the app store so it manages all your subscriptions so you pay an extra vig there to have it all nice and neat and not have to manage all those yeah maybe it's usually a little SPEAKER_29: bit easier to cancel like if you ever want to downgrade because you just do it you know through SPEAKER_239: the call on the phone like the wall street journal oh oh my god oh my god the wall street journal with SPEAKER_121: those dark times can you imagine being the sanctimonious i hate to some starting to sound like SPEAKER_243: trump okay sanctimonious fake news media uses dark patterns okay yeah nasty nasty no they do SPEAKER_247: on one side they're like criticizing everybody like the new york times and wall street journal SPEAKER_250: criticizing big tech and then they're doing dark patterns yeah you can you can sign up for anything you want you just can never leave and if you do want to leave you have to call a phone number and SPEAKER_24: they harangue you for 20 minutes it's also super confusing because they have like an app subscription and they have a ipad subscription and web subscription apple news yeah well no apple news is different SPEAKER_56: because some things are included in apple news and something it's all man it's all dark patterns SPEAKER_146: it's all dark patterns it's shame on you and then you know like i was having this issue with SPEAKER_13: peloton as well and tonal just let me pause my membership when i'm not using it i just want to SPEAKER_44: i think this is where lena khan could actually do something meaningful for consumers like ask consumers what's frustrating for them as opposed to like saying to consumers it's frustrating to have amazon prime it's not frustrating to have amazon prime you know what's frustrating not being SPEAKER_146: able to have a standard for pausing subscriptions canceling subscriptions make a standard to protect consumers that's it just like we did in banks where banks have to show you your interest rate SPEAKER_13: on the top level of your statement and they have to show you your interest paid to date or per year you know how you like all of a sudden that magically showed up in all your apps yeah did you notice that like when you open your app if you have like a margin loan or you have a mortgage it's like SPEAKER_145: very clear how much you paid an interest they got they forced them to that was like all dodd frank SPEAKER_13: i believe is that they were like listen be super clear about this you can't bury it or make it confusing you have to say here's how much interest you paid this year so you know people who were using their credit cards and getting charged 20 and they had 10 000 on their card they didn't realize they're paying like two thousand dollars a year you know to just float this ten thousand it kind of adds up and they go wow i'm an idiot i'm really paying a huge fig here or if it was like a 25 credit card SPEAKER_153: because they didn't have those introductory rates then the introductory rate would go down SPEAKER_262: and just everything was confusing six months and then it's like next thing you know it's crazy yeah SPEAKER_150: this is why i don't use any of this stuff listen selling software is hard enough right now man it's SPEAKER_153: hand-to-hand combat out there in b2b land the last thing you need to do is slow your sales team down because you don't have your sock 2 dialed in so if you're sas or a services company and you store consumer data in the cloud you know what you need to do you need to check out vanta they're going to get your sock 2 compliant easier and faster and vanta makes it so easy to get and renew your sock 2 on average vanta customers are sock 2 compliant just two to four weeks compare that to three to five months without vanta they're going to save you hundreds of hours of work and up to 85 on compliance costs and vanta does more than just sock 2. they also automate up to 90 compliance for gdpr hipaa and more you can't afford to lose out on those major customers the lighthouse customers the big fish the whales because of silly stuff like lacking compliance just work with vanta i'm an investor in the company it's a great company get your compliance automated get it tight tight is right and close those big deals here's the best part van is going to give you a thousand off because they love this weekend startups they love startups vanta.com twist that's v-a-n-t-a.com twist to get a thousand SPEAKER_268: dollars off your sock 2. what did you give your grade what did you give a b plus i give a solid SPEAKER_13: okay okay i mean listen i don't want people to think i'm in the bag uh here for elon because we're besties like i i honestly think i think it's fair a b out of the grade is incredible listen i would give the roadster a c as a sports car right it was like out of the gate and then you know his next thing the model s was an a plus so they have a i you know elon has a tendency to get something out there in the world get feedback and then nail it on the second try that that's my experience with his SPEAKER_275: products i think opposite of apple who waits and waits and waits waits and waits and then you're just SPEAKER_13: like okay great it's i'll be in your like i i used to be a big proponent of tiles yeah and then air tags came out and i was like yeah you know as much as i love my my tile team and i was team tile for a decade yeah i had to move to air tags because it's just integrated you know it's like people using like control 4 or questron or whatever yeah or savant and you eventually rip all that stuff out you throw it right in the garbage after you spend a quarter million dollars or a hundred thousand on your home automation and then use apple tv and sonos like i do yeah yeah why do you use that because kids your spouse your guests they all know how to use apple tv and sonos and then they can enjoy the music in SPEAKER_153: their room when they stay at your house or you know turn on the tv in the movie theater or turn on the tv SPEAKER_63: in the guest room or whatever just works it's just supposed to just work you're not supposed to be SPEAKER_13: fighting against this technology i feel like the gpt technology is getting there i had a little moment internally in the company people were writing emails that were not enthusiastic enough to founders and i'm like this is like very dry and boring i was like can you highlight this and go into notion ai grammarly ai claude bard chat gpt post your email to a founder in there and say David Friedberg: make this more enthusiastic yes and they did it and it worked and i'm like okay you are an incredibly boring writer yeah add more stoke add more stoke like chat and jt on that note let's get SPEAKER_24: into chat gpt because it actually has so we now have gpts last week we quickly demoed after demo day SPEAKER_59: uh definitely the assistant in the back end and now we have access to gpts and so what i'm going to do SPEAKER_218: here is go into oh let me get my screen share set up i was using the dolly one oh yeah it's technically a gpt SPEAKER_26: yeah so these are the ones made by open ai i've made some and uh you know we'll go through this whole process i actually made one for our bestie brad um but just what i want to highlight first is um you know so open ai is made a few different ones here um one of them is actually along kind of lines what you're saying is like this creative writing coach which helps you improve your writing skills um and you know math mentor mixologist laundry buddy i think there's a chef mentor classic gpt but i did something with uh game time which i can explain board games or card games player any age that the games SPEAKER_24: begin so i went in here and i i already have it kind of queued up so you know we don't have to do it in real time here so you asked it i asked it i have ace king diamond suit you know suited uh and the flop comes jack 10 9 with two diamonds what are my odds okay and basically it sounds like you're playing hold'em and the ace king of diamonds in your hand and the flop comes you know whatever i'm just reading it out here plot comes jack 10 9 with two of those being diamonds your head has SPEAKER_26: several potential strengths and it talks about my flush draw my straight draw my straight flush draw SPEAKER_23: and then combination odds which i think is you know pretty awesome if you're trying to learn some of SPEAKER_13: these games yes you're learning outs and it tells you here for the flush draw you have four diamonds needing one more for a flush okay so flush is five diamonds you just established that with folks there are 13 diamonds in the deck and you've seen four you are two plus the two on the board there are nine diamonds left out of the 47 unseen cards assuming a full table with no other diamonds no like somebody didn't flip over their diamonds the odds of hitting a diamond on the turn or approximately 19.1 percent nine out of 47 and the way you do that the easy way to do this back of the envelope math is to just give each of your outs times it by two two right yeah it's actually 2.1 SPEAKER_309: or something you get 18 yeah we do 18 on that right just if you did 2.1 you would round up and you'd be SPEAKER_13: like 19. so you got a one in essentially a one in five chance of turning it and then a one in five chance of rivering it which means you have a two and five chance which means you have a 40 chance of hitting yeah here straight draw you have a king and there's a jack and ten on the board so an open and a straight draw any queen or ace would give you a straight wait a second mean or ace would give you a SPEAKER_319: straight why would an ace give you a straight oh you said you have base 10 you got that wrong oh okay SPEAKER_11: bad advice a queen there are four queens so there are four queens and three remaining aces the aces SPEAKER_256: wouldn't give you a straight wow okay so you may not be ready for prime time uh somebody clip this and SPEAKER_328: send it to sam altman because he plays cards sam yeah here we go sam yeah i know sam likes to play cards SPEAKER_330: uh this is a hallucination that could question a lot of money yeah yeah yeah yeah and the straight SPEAKER_59: flush dry it messes that up a little bit too now that uh like i had just quickly done straight flush SPEAKER_17: raw specifically you were holding for the queen of no because i had only said i had said in the SPEAKER_24: original question jack ten nine comes with two of them being diamonds so i didn't clarify which SPEAKER_13: specify which two right if it had been ten jack of diamonds then a straight flush draw would be a possibility again who didn't it should have asked you a follow-up question which ones are the diamonds combined odds since you have both a flush draw and a straight draw you should consider the combined probability revolves a bit of math to avoid double counting the queen of diamonds right that's true 15 total outs nine diamonds plus six straight outs no you have three straight outs so it's actually SPEAKER_29: nine plus three is 12. um so maybe a disclaimer do not use this for your poker okay just yet if your SPEAKER_53: opponent is this yeah yeah and then we go all in what are my odds right and so basically i uh kind SPEAKER_218: of ran it through uh that but now we have to take this with a grain of salt because it you know these SPEAKER_264: details could be wrong our friends at corian provide wealth management services centered around you SPEAKER_153: corian's goals are to exceed your expectations simplify your life and help you establish a legacy that lasts for generations corian has been helping high achievers just like you enjoy their lives more fully preserve their wealth and provide for the people causes and communities they care about they're one of the largest integrated fee only u.s registered investment advisors and corian has deeply experienced teams in 23 strategic locations each team has extensive knowledge spanning the full spectrum of planning investing lending and money management disciplines the teams at corian put the collective power of their expertise into building you the custom wealth investment and family office solutions that can help you reach your holistic financial goals no matter how complex they may be SPEAKER_264: real wealth requires real solutions for more information speak with an advisor today at corian.com SPEAKER_346: that's corian.com can you show us how to make a gpt live on air yeah so i wanted to do one SPEAKER_24: for launch okay and um you know i made one for invest america let me just show you that one real SPEAKER_26: quickly okay and then and we're going to follow a similar process so this is the initiative that brad is uh working on brad gerson the fifth best exactly and you know you'll see here it has a url you can go to it and if i send this to you jcal and you open it on your phone i'll just text it to you real quickly neat thing that they've done just says something interesting is if i send this to you and you open that link it will open chat gpt but as invest america yes yeah so basically you get SPEAKER_98: dropped into the the invest america app gpts yeah are apps is a way to think of them and you can make your own apps exactly put them in the app store essentially or you just send the url to people so SPEAKER_13: if you wanted to do this again for shakespeare put shakespeare's combined works in there i guess as a pdf and you can yeah have people do that so here we go uh okay hold on i uh let me so with invest america SPEAKER_26: how did you do it what were they no so i just went to the website because i didn't have any docs and i was just doing it for fun and so um uh and i just said hey i clicked on tell me about invest america and it's going to go through a process here about you know all the content i've given it i can kind of SPEAKER_53: just we can look at it real quickly uh if i you know click here open it up and click edit gpt for some reason it's like running super slow right now so i don't know if it's like a popular yeah okay here SPEAKER_24: we go let's go here whoa okay and so if i go into this configuration you can see i gave it a name i SPEAKER_26: gave it a description i gave it some instructions okay which is this gpt provides information about SPEAKER_40: and i this is just a copy and paste from the website and i said the website is this the twitter account is SPEAKER_98: this so you're with the instructions you're being super clear that this gpt should focus on invest america SPEAKER_13: and giving it to resources so that it doesn't use its entire corpus it's narrowed things down SPEAKER_24: and it knows what you're talking about when you click on it great yeah and then i screenshotted a few uh you know things from the twitter account and gave it so that's all that that's all i did so SPEAKER_367: oh you gave screenshots of tweets from the twitter account very interesting so why don't we do one for SPEAKER_29: launch fund okay okay so let's just call it uh launch fund four got it uh this gp uh d provides information about launch fund four which you're an lpn yes i am okay and uh all right so let's let's say your first SPEAKER_297: capital call i assume you shipped it uh uh i have not shipped it yet but i got the capital SPEAKER_24: shippy-dippy-doo that day so i i scraped um a bunch of stuff from the website so i just put it in this just text document called the pdf document called launch fund four great right uh i also took a SPEAKER_34: couple of screenshots here okay this morning uh which should be on my desktop let me just make sure these are okay i think these are the two screenshots i took these are about the companies and the team SPEAKER_29: because i didn't have it anywhere else awesome all right and so let's just give it a quick instruction this uh gpt provides uh information uh about the new uh launch fund number four okay SPEAKER_385: anything else you want to put in here uh launch four is an early stage venture capital fund SPEAKER_387: okay uh managed by jason calacanis the host of this week in startups and the olin podcast so i'm just SPEAKER_121: trying to give it a little more context here so that maybe it doesn't confuse us because launch is a very generic name so it doesn't make a mistake there yeah okay but i don't know if that's necessary or not SPEAKER_370: yeah okay the fund is uh right here i'm just gonna grab this oops that's launch.co slash four yep okay SPEAKER_392: and uh maybe the twitter url do these are twitter urls we are we're at launch yeah our twitter url is SPEAKER_29: twitter.com slash launch there's x url is x.com slash launch okay and so we're done so basically we SPEAKER_53: do this we're going to save save and i'll just say only people oh we can have dolly generate a icon for it as well okay um but it'll just take a minute or two let's just do that so now i do that so it's like SPEAKER_183: making it live so i have the launch fund four okay so now i'm in this so if i if i text this to you SPEAKER_34: um just separately here i'm going to text it over to you jay cal so you have it you can try it yep but i'll do it here okay actually tell me about uh punch strategy the font strategy okay SPEAKER_26: and let's see what it comes up and so i just scraped the website gave it something and so here it goes SPEAKER_11: right okay number one early stage focus the fund is primarily targets c stage startups at their inception and supports them through series b funding rounds large portfolio with SPEAKER_13: significant ownership the fund aims to invest in over 400 early stage probably the goal of achieving 10 SPEAKER_403: ownership in the top 5 of these companies that's correct utilization of podcasts for reach and SPEAKER_277: discovery the fund uh leverages the popularity of jason calacanis's podcast correct founder university and launching salary the fund operates founder university launching salary finding versus an online course that helps founders ideas yeah this is great yeah i mean this is perfect like micro investments and future funding rounds for select startups graduating from the university SPEAKER_11: the fund makes friends and family style investments of 25 000 i mean this is just nailing it this is a SPEAKER_98: plus i mean this is a plus i would say if you're asking me to rate gpts out of the gate yeah i think it's an a SPEAKER_94: plus yeah i don't know what you think it's definitely an a so we're then we're debating a minus a or a plus SPEAKER_13: plus well can it get better of course it's going to get much better but is this going to change the game SPEAKER_75: of starting companies and doing information retrieval and creating services yes so the question is just SPEAKER_44: websites and websites the question is what's the time between when you could create a gpt slash app SPEAKER_75: or you could create one that replicates what slack does or what i don't know um create a new uber SPEAKER_51: create a a version of uh yeah you get the idea this is amazing yeah i mean it's it's like and look it SPEAKER_24: just took a you know five minutes right i had to scrape that's the thing that's incredible about this SPEAKER_261: is how quickly you can do it yeah and imagine this incorporates your stripe account right yes uh in the SPEAKER_26: next version imagine it allows you to connect it can you can like we just we made a we made like an informational um gpt yeah there was other features in there that allow you to you know so if i go back into this uh interface here here you know if we go down below you can add these actions and you can start basically connecting it to other services that you have so okay we've made a very simplistic one SPEAKER_420: here um which is imagine this was open table and you could book a restaurant or yelp made one and you SPEAKER_421: could yeah query the reviews and then have it actually make the reservation for you so if you SPEAKER_277: start thinking about these as agents that go out into the world like baby gbt we talked about when we started doing this like that's kind of the next step is what can it do once it gives you the answer like yeah what if it had on there would you like to book a call with jason calacanis to talk about this then it asks you three questions yeah have you invested in a venture capital firm before do you have any questions about venture and it tried to answer them you get the idea well you you can do SPEAKER_26: that so we will add that in that jason you know sometimes you really are the world's greatest moderator because you hit it so we are going thank you for saying that according to the youtube comments SPEAKER_429: some people prefer other moderators who maybe don't ask as challenging of questions to certain guests SPEAKER_432: say interest interested uh in investing uh direct that to this europe okay okay so um we just we added SPEAKER_53: that and i was using dolly to come up with uh an icon here but it's taking a this thing is a bit slow SPEAKER_261: i don't know it kind of takes uh dolly seems like it's really resource intensive yeah that always happens to me dolly takes way too long i think it's like amongst their most uh intensive and it probably SPEAKER_13: cost them a ton of money to do dolly so they probably make it slow and put it on like a slower SPEAKER_40: framework so yeah you know it doesn't cost so i just added that so i said i'm interested in investing SPEAKER_372: let's see if it incorporated that actual um piece i just put in there if it didn't but um SPEAKER_36: um there you go right and so there it is we just did that real so it's all of this is happening real SPEAKER_277: time instantaneously it doesn't require a developer yeah yeah yeah um and you could put in there hey if somebody wants to do a meeting please ask them these five questions to qualify them and then yes you know send them to this calendar and give them three dates to choose from so you don't even have to SPEAKER_421: go to the type form like we're sending you a type form it could do that whole interaction with you yeah it changes everything this is changes everything really shifting yeah yeah it's totally paradigm SPEAKER_13: shifting and apple and android and you know people who have app stores need to look at this so this is SPEAKER_456: a message to tim cook this is a message to sundar sundar this is the next app store you don't have it SPEAKER_13: fire 100 people bring 200 people into a room fire 100 of them and then the other 100 put in charge and do like a darth vader type thing like this has to be up and running within 120 days and has to be better it has to be not only feature just you 200 people have 120 days to make this better and if it's not SPEAKER_421: better and then fire 100 people in the room and then put the other 100 in charge yeah you got to take SPEAKER_346: a like a darth vader approach to this because you're not you're moving way too slow and imagine SPEAKER_24: like you know so much of our interaction is on text and um you know you if someone asks you like oh jason tell me more about your fund and you just send them that and it opens up like open ai but in you know SPEAKER_53: the launch fund 4 view and like i just chat with it like how powerful is that going to be crazy the SPEAKER_75: whole thing's good and then if it had a human seeing the chat and fixing it in real time so let's say every time it fires up one of these chats we get a slack room like a throwaway slack room that just shows them interacting and we're kind of watching it occur and a human can jump in if it's wrong and you can highlight it and say this is the correct answer for next time or you could say human coming in to intervene so this is going to get super interesting imagine it's answering questions for american airlines and there's an american airlines gpt and you just start talking to like why doesn't american airlines have one of these yeah yeah but i mean imagine it's answering questions about SPEAKER_13: upgrading the lounge and then it's got something wrong about the lounge right and it's like oh if you have a first class ticket you can a business class ticket you can go into the lounge that's not always true sometimes your business they're like yeah you can't use the lounge got to be SPEAKER_40: international business now to get it or something yeah and so it's like okay you can't use the lounge SPEAKER_13: it's like one of the stupidest craziest thing right but just make it simple if you're in business class you get the lounge if you're not in business class you don't get the lounge why does this have to be like a game every time i go up there like i'm some pauper because they gave they gave too SPEAKER_53: many people access to the lounge with credit cards that now they have to claw it back okay it's not SPEAKER_13: my problem yeah just if i bought a business class ticket let me go have a cup of bad coffee and use your slightly better wi-fi and sit in a less hard chair yeah nicer bathroom with that yeah with a bathroom that's been cleaned more than once every three days okay great but don't make it complicated anyway if you were watching that and it did have something there it could highlight and say to check if your business class ticket is go to the front counter sometimes it is sometimes it and SPEAKER_75: then you just fix that or human could jump in and say you're at jfk we're actually under construction it's not available until january 1st 2024 and then it incorporates that into the next answer so imagine i'm watching so i'm a customer support rep i'm watching sandeep 10 of these in real time and i'm just sitting there like this watching that one's wrong boom i go to number two i fix it boom this one's wrong now imagine i'm doing that not for customer service at american airlines but i'm doing that for uh giving advice on indigestion so i'm a doctor i'm an emergency care yep uh and i see people coming in and saying have indigestion and ask them this question hey what did you eat when did you last eat when did this start have you vomited have you done this and then it says oh wait a second this person says they um ate some liquid soap or something and it's oh yeah go to the emergency room now you know like call 9-1-1 and can you imagine if that was done by ai how many lives would be saved now listen we just did the one with uh doing poker odds and it was wrong so you would SPEAKER_146: you would want to be careful with this you would be monitoring 9-11 calls in countries where you don't have access to a doctor so the gpt would be by default better than nothing and then you have some human watching it fixing in real time there are emerging markets where probably people don't SPEAKER_75: have access to doctors and they're looking stuff up on their phone or they're just asking people you know for word of mouth recommendations for an upset stomach they don't you know know what 20 questions to ask this i think vinoca was right i think these things would do a better job you tell me what you SPEAKER_146: think for the global population than doctors would or because they don't have access to a doctor or if you did have access to doctor i wonder if this would be better because it would be lower cost i don't SPEAKER_26: know how close we are two examples so i think there was a woman who put in um you know like the latest version of chat gpt you can you know especially after the dev day launch everyone has access where you can just upload like a pdf and it's not like code interpreter but just analyze it for you and then SPEAKER_24: she uploaded like her blood results and it basically highlighted something and it said uh you know she had this problem for a long time and it just had a different view and she's like oh my god thanks for doing that because like nobody ever did that and then me personally um you know my dad has parkinson's and so he's on all these meds and the other day is it's like i literally just took a picture of his uh subscription and just put it in the chat to me and tell me what all these drugs are and it just did it for me and so yes rather than typing them all out and some are generics SPEAKER_26: and some are not so it's a game changer it's a real real disclaimer disclaimer disclaimer the stuff SPEAKER_492: hallucinates it gets things wrong so you want to check but as a first line in your discovery i don't see any harm to having it tell you what is in your blood test and then giving that response to your SPEAKER_75: doctor and saying hey this thing thinks my iron deficiency you know i should have this treatment and this thing what do you think i don't know where it's getting this information on the open web SPEAKER_73: from but i mean it sounds to me like every single time you get any kind of a doctor report or blood thing there's no downside to putting it through there and getting a third opinion and SPEAKER_157: then bringing it back to your doctor and doing it in consultation with them well i mean it's amazing SPEAKER_26: it's like it's i really like that idea um like you said you know we always go for a second or third opinion why can't you get one um uh and i think this is the article i'll just pull it up here and uh SPEAKER_40: yeah so i think it says like woman claims had gb found correct diagnosis for her son's chronic pain she put it in there and then basically had these meltdowns and it found out so i think it's excellent SPEAKER_59: yeah really really good oh man i'm getting these ads running on here now that's awesome that's why the SPEAKER_53: internet's gonna go away too because uh these ads are everywhere now and it's we're gonna have Chamath Palihapitiya: a different model it's much cleaner now uh now we do need to get into the discussion of copyright SPEAKER_13: and now that there's a business being built out of this yeah who gets paid because if you are giving this SPEAKER_456: advice uh and it's been trained on webmd or other commercial enterprises data hmm there's an argument like can i get on the other side i'm saying what if they got textbooks textbooks SPEAKER_13: those textbooks trust me those are some of the most insane uh but they could have licensed them we SPEAKER_94: don't know that i don't think oh i have not heard of open ai paying a single license have you well many Chamath Palihapitiya: textbooks are out of copyright right what's the copyright uh you can keep um you can keep if you update them you can keep them perpetually under so if you were to like just change a sentence you could SPEAKER_13: then renew it oh really uh yeah so it's you're talking about artworks like a book like great gatsby or something like that and those can be i think it's 75 years or 50 years past the life of uh you just asked chat gpt what's the copyright on a book but i think it's 50 years after the author dies um and then it goes into public domain but they keep increasing that and they call it the mickey mouse rule so i think they made it 75 after the life of the thing because of when roy disney died they wanted to add 75 years to that to protect the disney family so they call it the disney rule now but yeah here we go the copyright duration for textbooks the united states works published in 1976 it lasts the life of the author plus 70 years exactly what i said for works created for hire anonymous or pseudonym works the duration is 95 years of the publication 120 years creation which ever is shorter european year in general is life of the author plus 70 years uh copyright laws can vary SPEAKER_387: significantly most countries life plus 70 years rule okay wow interesting so you learn something new SPEAKER_522: every day on chat gpt yeah there you do we do all right uh do we want to do one more i think we have time for one more if you add one or we can just wrap up it's up to you okay we'll do one more we'll do SPEAKER_26: one more right we'll give them what they want we give them what they want okay so this let me get this set up uh and i i i actually found this and want to use it myself and so let me go here um this is called SPEAKER_24: fake spot reviews analysis and so this was like a historical plugin that you could add to your chrome SPEAKER_26: and you can see it here i have it uh added in and on any um you know a couple of different sites i'm just on amazon here for those listening you can see here i can say analyze reviews of the you know SPEAKER_53: third generation airpods so you click this analyze reviews okay and so it's like a chrome extension you SPEAKER_29: put in it's a it's a chrome extension got it and you know before it would do these pros and cons but SPEAKER_26: now it has this little chat bot here and this chat bot lets you chat with all the reviews i like that SPEAKER_527: and so uh are the air uh pros good for large ears let's just say okay stupid question but basically you SPEAKER_26: know like someone may have and uh um and there you go right so it's it's kind of looked through all the SPEAKER_421: reviews it can give you an answer i think that's like super no this is super helpful because i am often trying to being a product wizard myself yeah ask very nuanced questions and yeah it's very hard to SPEAKER_13: get through all the things but i know how to search there is a search feature in the amazon reviews but SPEAKER_98: i oh there is okay i didn't have a search feature but they don't have a chatbot okay and so you're searching for keywords and so you can search for battery or yeah whatever you're you're looking for you know if you're looking for the charging speed or whatever you know so yeah fantastic awesome SPEAKER_13: yeah uh so i give that a b that's a that's a b for me okay um all right you know it's nothing spectacular nothing bad i give it a b minus like i would have liked to see something more interesting SPEAKER_387: like it um and so i encourage people making it to make it more than i would expect i don't just want to chatbot that's going to be like easy they could turn that on in 10 seconds yeah you know really easily turn on 10 seconds so um i would go with something more unique what would be a unique feature SPEAKER_13: to do with the reviews maybe tell me sentiment of the reviews over time yeah so how how many months of reviews are there and what were the major um what were the major comments they actually they don't have SPEAKER_26: quite that but they have like uh you know it has this many reviews and yeah it has some some amount of SPEAKER_547: that i'd like to tell me if there's fake reviews too i mean i think it's cool i'll be honest yeah SPEAKER_26: it says it says amazon rating is this but fake spot adjusted rating is this SPEAKER_550: oh yeah yeah oh that i like okay i'm gonna go b i'm gonna go b plus maybe even yeah because i that's SPEAKER_13: a that's an acute problem i have i think when you're tackling this you always got to think about you know um you got to think about what are the unique problems or frustrations people have and the unique problem frustrations with reviews is how fake they are yeah or how fake they seem SPEAKER_26: and so that's something we could all use help with yeah all right i'm kind of with you b plus i think is good you can always make it better um let's do it yeah well also i think back to anticipating if David Friedberg: whatever question i'm asking there you ask the question are they good for large years SPEAKER_98: it should be asking what other airpod competitors are good for large years or which so you should whatever query somebody does you should have another gpt running that says what other questions should i SPEAKER_75: ask this cons this potential purchaser in order to help them solve their problem correct and then show SPEAKER_98: you those three purchase yeah drive you should be um oh is this is this for you or for a loved one SPEAKER_146: or yeah um and how large are your ears have you ever tried other have you do you have air do you have any ear ones currently that work well which brand are they or have you considered these three other brands we hear that they have you know um it turns out the ones from google that i like the pixel buds yeah i like the pixel buds because they're flat flat yeah so when you lay on your pillow at night with your airpods you you ever get that you ever try to lay on your pillow with airpods oh yeah yeah you wake up it hurts yeah yeah okay hey uh producer john producer court do me a favor uh will you send uh mr sandeep a set of google pixel buds from the fine folks at this week in SPEAKER_565: startups so say thank you okay that's mr sandeep sorry i say thank you to mr sandeep SPEAKER_568: i'm trying to teach young kids the manners here you guys see mr sandeep is here yeah uh i like referring to people by their last name it's so jarring so jarring to call mr calacanis SPEAKER_572: yeah yeah yeah did you have that happening now that people are like who work for you or who work in SPEAKER_53: your orbit call you mr calacanis only when you like travel like you know in like certain countries and you get it right like oh man i'm like don't i'm like i turn around at least you must have had it SPEAKER_577: i'm like is my dad here did my dad come with me yeah mr calacanis right it's been another amazing Chamath Palihapitiya: episode of this week in startups with your guy sandeep madra if you would like uh go to definitive.io SPEAKER_13: and uh you can hire definitive intelligence to do partnerships and consultations but i think they're dance cards full so if you're a large company and you're spending uh you know let's say mid figure mid six figures engagement to solve your ai problems his team is the one that will solve SPEAKER_53: them very quick um that dashboard that you wanted the llm ops dashboard will do it for you oh yeah SPEAKER_43: what wait what does that just remind me what we want to do no so like that you know what we were talking about there is called llm ops like llm operations so i think a lot of businesses now are SPEAKER_29: focusing on commercializing you know llm applications and you need like operations around SPEAKER_40: it so that you can do those kind of things and so yeah for shopping with a lot of that infrastructure SPEAKER_261: ah got it yeah that just the infrastructure here and the training of teams is going to take a couple SPEAKER_13: years to get this right uh and definitive.io will do it for you faster and you're in a race folks so get on the horn with sunny he's sun deep on twitter slash x formerly known as twitter and SPEAKER_301: we'll see you all next time on this week in startups hey write a review if you like the show talk to you later bye