Navigating the AI Frontier in 2023: Lessons, Pivots and Triumphs from Two Early-stage Founders

Hi welcome uh it's great to be here Today my name is Brendan Kim I'm with Samsung next I'm one of the managing Directors there Samsung next is a Investment arm for Samsung electronics And we invest in Founders Um in areas that Samsung cares about and Of course AI is all the rage and Samsung Within Samsung we're very much Interested in what's happening there so We're making a lot of early investments In Ai and we have today two companies in Our portfolio some of the most promising Companies in our portfolio I want to Introduce them and give them a chance to Talk about themselves a little bit and Maybe introduce the company I'll start With vike as in Viking right so Viking Yeah bike go ahead hi all I'm Mike CEO Of Dynamo FL uh background about myself I got my masters and PhD from mit's Computer science department and privacy Preserving machine learning and Basically converted my thesis to my Startup Dynamo FL so what do we do at Dynamo FL we offer the end-to-end Infrastructure for training privacy Preserving and regulation compliant Generative AI models so everything from Like identifying all the Privacy Vulnerabilities in your large language Models or computer vision models to Making sure how do you train them in an Optimized and in a regulation compliant

Manner to effectively deploying them on Like device or like servers we provide The end-to-end infrastructure And then also with us is Yana so that's Like Anna but with a Yana that's right Yeah I'm Jana willinger Um CEO and founder of craftful craftful Is a co-pilots for product managers and Other product Builders my background Before starting craft fall is that I was A product manager and ultimately LED Product teams at various tech companies And we always struggled with all the User feedback we were getting from Channels like app store reviews support Tickets various sales calls even Twitter And so as I started experimenting with Large language models this was obviously A use case that I tried out we ended up Pivoting to this current solution at the Beginning of the year have now over 20 000 users use craftful Every day to analyze all of their user Feedback and incorporate it into their Product development process they've now Saved over a hundred thousand hours Which means that they can use that time For strategic work which is really Really exciting that's what I would have Wanted to do as a PM fantastic thanks Um I think this is a really interesting To talk about like when the first uh Public launch of chat GPT came out There's this huge enthusiasm for a gen

Ai and it seemed like you could disrupt Anything and everything like you could Start a startup in gen Ai and everything Was up for grabs I think we're coming Into the realization that maybe that's Not really the case that that yes AI has Lowered the costs of innovation and it Can be very disruptive but like any Other space it's not like you can just Pick a spot and just just just do it you Have to really like figure out where you Want to focus right and so if I can Wonder if you can tell us a little bit About like why uh Dynamo FL how it's Found its spot and Um you know what it took to get there Yeah definitely so our Focus from day One has always been in privacy so um When like Chad GPT came it was actually To our advantage because uh one drawback Of Chad GPT as well because you guys Were you guys had started before chat GPT yeah we started like yeah way before Like a year before um chat GPT so uh uh Before that it was called like more like Natural language processing and when Jtbd came the word llms became famous Right um right but like what iuc is like Uh obviously as people start using chat GPD big Financial companies or insurance Companies they were obviously facing an Obvious challenge which is like you just Can't send like say sensitive data or Like sensitive documents to a

Third-party provider because it has a Lot of sensitive information so Basically it was just like a blessing in Disguise for us uh where like our like Entire privacy infrastructure we just Had to like make sure that we rebranded Our company as like more of a privacy Preserving llm company um because it was Much easier for us to like sell to our Big Fortune 500 Enterprises for whom This was actually a big challenge yeah So um yeah I think like uh it was Disruptive but in a good way yeah it was For you as you guys were sort of well Positioned when this all came about Brianna you weren't you you had to like Pivot yourself into this right so what What was your thought process and and Like picking that spot picking the spot Around you know where where you're Operating yeah absolutely so we're Building a Um a different solution for for product Managers and product Builders back then Um and uh we did have Um actually way before Chachi PT came Out we had a gpt3 feature uh powered Feature in in our product and at some Point we realized that actually this Feature is taking over our product where We're getting much more interest for That feature specifically we're getting It outside of the specific Um subset of product managers that we

Were we were targeting and so that that Was sort of like my first clue even Though it's a prop it's a problem that I've experienced myself as as a as a Product manager I still went out and Validated it before before doubling down On it Um so I you know went went to a Conference around the same time last Year Um there's the masters of skills Summit Invite only conference it's a good Audience to Target and I was basically Just introduce myself and be like I'm Jana CEO of craftful we do this and then I would listen for feedback to see what Am I getting what kind of feedback am I Getting from product Builders what sort Of like solutions would they want to see Uh built with this right like they they Assumed we were already that that Pivoted product at that point and so That's I kind of leveraged that to be Our positioning to be that initial uh You know feature set that we pivoted With and then we went from there that That was sort of like our our starting Point and was that like a a natural Pivot for you or because sometimes when You pivot you're like you don't want to Leave the other stuff behind right you Want to cling to where what you what you Know what you started with and it can be Hard sometimes to let all that go and

And grab something that you you think Might be it but you're not quite sure Yet either right yeah I think it just Took off so incredibly fast that we none Of us on the team had had any time to Just reflect on it we you know we uh Changed changed our positioning started Building this new feature set and while We're doing that people were organically Just signing up via Twitter just you Know like following a lot like you know Tweets random tweets about non-alcoholic Beer and people would come and start Signing up for a product just because They saw what what our landing page said So we never really had a chance to think About what we're leaving behind because We just felt this pool to where we were Going a lot of pool and so you knew that Was where you need to run yeah that's Fantastic Um you know one of the things that uh Struck me like is Um You know you're talking about privacy And uh you know when you when you enter A hype cycle uh like an AI Um you tend to want to adopt all the Language of you know that hype cycle and You know you're everybody that used to Be something else is now an AI company Everybody that used to be something else Is now a privacy company everybody that Used to be something else is now you

Know some kind of data you know private You know proprietary data company right So um As you you mentioned you you kind of Changed your language a bit right and so You know what did you have to watch out For what did you do right and what do You feel like you maybe you didn't do Right or you know what do you have to Watch for yeah I I to answer that Question specifically I think what we Did right was to make sure that like say For example our landing page we just Literally had to like say that ah this Is the ultimate infrastructure for Training privacy preserving large Language models we would identify all Your privacy issues and um it also it's Also the market which I think like Played a key role because someone like a Bloomberg when they released their Bloomberg GPT what we saw was they Highlighted that yes we are not Releasing the model uh the main reason Being that like people can't like Extract sensitive pis or information From this model so like we didn't even Have to do that publicity we had like Companies who were like openly accepting That there are so many like privacy Vulnerabilities so all we had to do was Like we wrote like blogs saying that how You can extract information from say Like chat GPT or like other kind kind of

Like language models and so basically From projecting ourselves as like an Like a machine learning optimization With privacy company it was more like Privacy and regulation compliance and Yeah we just had to like literally uh Brand ourselves that way and literally In like three weeks we were able to get Our series a done so um like the timing Was perfect the timing was perfect for Your round yeah and I think like it's Basically like we just got to know what The market like people like wanted and We were early enough to like Um position ourselves and like Rebrand Ourselves uh in a way which like we knew That this is a software which big Fortune hundred companies 500s are Definitely going to use Um yeah And Jana for you when you sort of did This pivot were there any raised Eyebrows like wait kraffle is doing that Now it's like that wasn't quite what you Were talking about before and so how did You how did you kind of manage that Process Um I mean I think you know we went from Something where we didn't have As as much usage to something where we Had a bunch of usage quickly so I think We were just known for very quickly Became known for what for for the Pivoted version of us so that wasn't

Necessarily as as big of a question mark I think we we did Um Uh you know we did time things right in Terms of we always kind of tried to Build a very to begin with a very quick Very limited features so that we could Ride the wave of everything that was Happening so we launched our um uh kind Of a very first MVP right before the Chat GPT API launched which was right Before the gpt4 launched and so I think We could write out this uh wave where Folks were product managers and product Builders wanted to use this new Technology for their workflow and we Were there and we had a solution for Them so I think just having that and Being able to talk like we were like Chat GPT for product Builders and then Over time evolved to to our current Positioning as co-pilots but I think Just being there having a very specific Solution for the specific audience while Leveraging everything that else that was Sort of going on Um I think was sort of it was a helpful Use of hypewave yeah yeah do either of You have any thoughts about like you Know um you have all these buzzwords Right and do you just slap everything on Or and see what sticks or do you are you Do you have to be more careful do you You know how do you do you find a

Balance I think it's more uh less so From a positioning perspective but more On the like are people constantly using Our products uh when I talk with users Are they telling me here's how your Solution solves my problem and here's How I used to do you know like I would Get on in the beginning I would get on Calls with like a piece of product and They would they would screen share and Show me how they manually were doing the Same thing that our product did but it Took them 100 you know took them hours To do it over the weekend but now they Can just do it in minutes so just like The sense of like am I actually solving A problem I think in hype wave if the Feedback you're getting is something Like oh this is a really cool Solution That's then you know I've used the hype Terminology and everyone's hype excited About it but there's not there's nothing There right that's that's what you want To avoid yeah yeah Um Both of you I think are talking a lot About like that product Market fit that You're finding and I'm wondering what uh If there are any particular metrics that That the two of you look at Um you know what how do you know when You've started to find that product Market fit like what scale are you are We talking about

Um how do you how do you get that sense Yeah for us it was mainly like uh I mean We we haven't hired any sales people in Our team till date it's all been like Inbound from like we're talking about Like some of the top most like Consulting companies finance companies Insurance companies so I feel that like That is what showed us that okay we've Actually hit product Market fit because Even without any like marketing we were Getting inbound interest from like the Cxos and like the the VPS from these Like big like Fortune 50 1400 companies And um one thing we were seeing were That like the the rate at which say for Example onboarding an Enterprise Customer Um We were doing it like say for example Like back in the days if it was like one Uh one big customer every month now we Are talking about like one customer Every two weeks and now it's at the Place where like we literally have like Some of the largest companies really Onboarding once every week so I think That is when we realized we've hit Product Market fit because this has Become such a key problem without which Like literally organizations cannot Deploy any generative applications uh so I think that's the moment we realized That you know pmf is how about for you

Jana what's yeah so we primarily look at Adoption and stickiness in turn as our Kind of key Um obviously we're not we don't have any Like natural virality in our products so It's been looking at when people come to Our landing page do they instantly want To use the product right are they Signing up and are they inviting the Rest of their team that's sort of like The how many invites are are users Sending along and then kind of the the Key pieces are they using us you know Every day or every week whenever they're Like Cycles uh make sense for their Product building are we seeing that Stickiness for the folks who have done The right kind you know who have Connected their their feedback and Actually Should be getting value out of the Product we are also looking at churn Just to get a sense for like you know if If books turn why are they churning is There something that that's uh that Isn't you know that we can optimize but That's less around product Market fit And more just around product Optimization You mentioned earlier that there was a Lot of pull right and you're talking About hey there's you're getting a lot Of inbound it's not much outbound is That what you're looking for is that

Sense of pull yeah absolutely all of um All of all of our usage at this point This 100 organic inbound Um so it's which originally uh came Mostly from Twitter after after some Point we started getting featured by Product managers and various products Um uh blogs and newsletters podcasts and So it's sort of like went snowballed From there but it's been it's been 100 Organic at this point then and like You're doing the same thing it sounds Like it's not really active marketing or We're pushing sales it's it's funny Exactly we just write like blogs Put out stuff there and then like Organic inbound yeah yeah okay Um but it is getting your message out in Some way right whether it's through Frogs or through some kind of Newsletters and things like that Um You know you know I think the this area That that both of you are in in this Whole AI space is like Changing so quickly and and moving so Fast and Um How do you keep up like what do you what Do you guys do to keep up keep on top of The latest stuff right yeah well for us Like being like a deep tech company we Have to like make sure like uh we we Constantly like hire the best people and

Like make sure like because a lot of Like are the stuff which we do is like Research yeah right like um hardcore Like llm optimizations or like privacy Research so Um I think like uh in order to be top It's really important to be like up to Date in all the technology not only Thinking about what is needed say like Right now for the customers but also Like kind of like anticipating what is Needed like in a year yeah um so for you Like I mean privacy is uh such a Well-known issue especially for Enterprises that seems like everybody Says that there can protect data and Make things private right so how do you Keep above the noise and like you know Uh I mean again like this is what we say Right like this is not something which We just came up with in the last like Say a year like or like last like six Months once chat GPD became famous right Like a lot of us in our team we all have Like our phds like like experience in Like developing like privacy preserving Machine learning protocols for like six Seven years yeah so kind of like this Gets reflected whenever we talk to a Customer because the customers are also Like very well educated when you're Talking about big Enterprise companies So they exactly know what they want so Um I think that's how you differentiate

From the noises like you actually show That like not only your product is easy To use but you are also an expert in the Space and can actually help the customer Like navigate all the problems and Issues which they're facing yeah Yana For you how do you how does craftful Stay on top of things respond quickly I Mean because it is constantly changing Space and now you're hearing co-pilots Right co-pilot for everything right so That's right yeah I mean I think the Technology is changing really rapidly And I think there's just a lot of Confusion in the space generally I did Um earlier uh earlier this year I did Start a community for Gen AI Founders And it's now a few hundred Founders Where we exchange advice on on kind of How to build in the space and pitch in The space and physician in this space Um so that that's that's been incredibly Valuable I actually we're a white Combinator company and I think Obviously normally we just lean on the y Combinator network when when I started The gmai community it was that was in The very beginning of this by now a Bunch of YC companies are in the same Space so now I feel like I'm getting as Much information from both both of those Founder communities I would say that That's kind of that's been a really big Part for staying up to date is just

Talking with Founders that are all Building in the space and are Experimenting for example at one point Earlier this year we sort of we hit the Limit we outgrew gpt4 because we just Had too much data that we were sending In and the token per limit uh limit was Limiting how much data we could send in And analyze per day and so I went out to To my network and started asking kind of Like what has anyone come across this And actually we there was a certain rite Of passage like there wasn't a whole lot Of folks who had this problem and this Is not because we had so many users back Then we just had so much data yeah um And and I got really great advice from From the community in terms of have you Tried different different things I Wanted to just follow up on that and This might end up being the last Question but Um you're getting a lot of thoughts from The community Um you know uh everybody seems to be an Expert in AI these days and so they're All bombarding with I just try to just Try that I used to be a a blockchain Investor a couple years ago and I'm now I'm an AI investor right so uh how do You manage that kind of feedback and as A builder uh what what what what kinds Of advice do you look for what kinds of Advice do you think yeah you know not so

Much I mean I think like the Question because as a Founder your time Is so so limited as the founder at a Fast growing company your time is Incredibly limited you just you have you Just have a second to think about Everything basically so you have to Think about what's the problem that I'm Trying to figure out and who is the best Person in my network to be focused Exactly like you don't want to listen to Everyone you like and even unless I go Out to the commute to the general AI Community I'm looking at who's building These kind of LM applications who've had Scale who you know to be focused and be Targeted about about what you're looking For how about you like um I think like Listen to the customer so like customer Customer like demands like at the end of The day you're building a business and Like forget the investors forget the Advice yeah customers customers put them First make sure you're always like Available for them and just like focus On their like problems yeah well 20 Minutes went by really fast Um so I really want to thank everybody Here for attending and to TechCrunch for Putting all this together uh to bike and Yana thank you so much we really Appreciate your time here thank you yeah Thank you for having us

Coinbase
OUR TAKE

Coinbase is a popular cryptocurrency exchange. It makes it easy to buy, sell, and exchange cryptocurrencies like Bitcoin. Coinbase also has a brokerage service that makes it easy to buy Bitcoin as easily as buying stocks through an online broker. However, Coinbase can be expensive due to the fees it charges and its poor customer service.

Leave a Comment

    • bitcoinBitcoin (BTC) $ 69,672.00 4.76%
    • ethereumEthereum (ETH) $ 3,611.77 4.88%
    • tetherTether (USDT) $ 1.00 0.09%
    • bnbBNB (BNB) $ 631.25 5.34%
    • solanaSolana (SOL) $ 158.75 8.25%
    • staked-etherLido Staked Ether (STETH) $ 3,610.74 4.87%
    • usd-coinUSDC (USDC) $ 1.00 0.08%
    • xrpXRP (XRP) $ 0.494667 4.27%
    • dogecoinDogecoin (DOGE) $ 0.149397 10.9%
    • the-open-networkToncoin (TON) $ 7.54 10.29%