Latest / Newcomer Pod / Databricks CEO Ali Ghodsi & MosaicML Founder Naveen Rao Speak at Cerebral Valley
Transcript
- 0:00Hey, it's Eric Newcomer. Welcome to the Newcomer Podcast.
- 0:03We've got a great episode for you this week.
- 0:05Coming to you live from the Cerebral Valley AI Summit, I
- 0:09spoke with Databrick CEO Ali Godsey and Mosaic founder Naveen
- 0:14Rao. The duo met at the first
- 0:17Cerebral Valley and eventually Ali acquired Naveen's company
- 0:21for $1.3 billion. As we talk about on stage, this
- 0:24conversation took place November 15th, but I think it's still
- 0:29very relevant to you. So give it a listen before we
- 0:32get to that conversation. A word from our sponsors, Oracle
- 0:36and NVIDIA. Thousands of enterprises around
- 0:39the world rely on Oracle Cloud Infrastructure, OCI to power
- 0:43applications that drive their businesses.
- 0:45OCI customers include leaders across industries such as
- 0:48healthcare, scientific research, financial services,
- 0:51telecommunications and more. OCI also works with NVIDIA to
- 0:56provide an AI training as a service platform for customers
- 0:59to train complex AI models. Talk with Oracle about
- 1:02accelerating your GPU workloads at the link in the description.
- 1:06And now here's my conversation with Databrick CEO Ali Godsey
- 1:11and Mosaic founder Naveen Rao. Welcome.
- 1:14Thank you so much. You guys have been such big
- 1:16supporters of the conference. Thanks for coming back and
- 1:18getting on stage this time. Tell tell the story to start off
- 1:21with just of the acquisition. You know, I know we we bragged
- 1:25many times about how you met at Cerebral Valley, but how does
- 1:28that turn into an actual merger or acquisition?
- 1:32We met during the cerebral valley.
- 1:34But then you had this thing in the evening.
- 1:36What was that? There was like a a speaker
- 1:38dinner. Yeah, you were there.
- 1:39Speaker dinner and yeah, it was like 8-9 PM and that's when we
- 1:42started like hanging out and talking about, hey, how's it
- 1:44going? What are you doing?
- 1:45You were telling me war stories, but I don't think I'm allowed
- 1:47to, yeah. Please don't repeat those.
- 1:49Keep them between us. And yeah, so started talking to
- 1:51Naveen about what he's doing. And I was like, oh, he seems
- 1:53pretty legit, like, you know, and the business is pretty good
- 1:55and, you know, so I started looking into more and more
- 1:58Mosaic ML and what they're doing and, you know, everything
- 2:02looked, you know, super awesome. And I I remember like, God,
- 2:04these guys are pretty good. And then I walk out of my office
- 2:07in SF and I see one of our key employees, Shangri.
- 2:11He's sitting there on the laptop and I look at, I look over his
- 2:12shoulder, see what he's doing. Hopefully he's working and you
- 2:15know, and he's on the Mosaic ML website and I'm like, hey, what
- 2:18are you looking at? And he's like, hey, these guys
- 2:20just released, like, and then you're just released serving
- 2:22model serving of, you know, GPUs.
- 2:24And I was like, wow, what? What's is it any good?
- 2:27He's like, yeah, this is very competitive pricing.
- 2:28This will be difficult. Like this is pretty good.
- 2:30And I was like, huh, So like, all the stars are lining.
- 2:33And that day I called the guy who runs all of like,
- 2:36acquisitions and so on for Native bricks.
- 2:38And I said hey, like this company I just saw yesterday.
- 2:40And they seem pretty good. And he's like, Naveen at Mosaic.
- 2:43I was like, yeah, yeah, yeah. He's like if you want that
- 2:46company, you have to call him this weekend and by this
- 2:49weekend. Yeah.
- 2:50When you get that call, like what's what's your thought
- 2:52process you you build this a start up for the long term like
- 2:56yeah, what what were you thinking when you got the call?
- 2:58Well, you know, it's using the dating analogy, right?
- 3:01It's like you kind of know he's going to pop the question, you
- 3:03know? How do I know it?
- 3:06Smirk on his face as I called him.
- 3:07You could see him. He was like, sitting there.
- 3:09Like he knows. And this had happened to you
- 3:10before your last. Yeah.
- 3:12And it always happens faster than people think.
- 3:14It's like one of these things, like once everything aligns,
- 3:16it's it's like go, you made the decision go, right.
- 3:19So I kind of knew what I was going to think through a little
- 3:22bit, but it wasn't a no ally. I don't want to do this.
- 3:25It was like, no, no, no, this actually makes a lot of sense.
- 3:27Let's talk about how that could work.
- 3:29And basically, I think in the first conversation, I even said,
- 3:31like, I think it's going to come down to economics.
- 3:33Can we make this work? I'll quote you exactly.
- 3:35You said this makes sense. How much of the upside am I
- 3:37getting? It was the exact.
- 3:39Quote it was like. Depends on how much of the
- 3:42upside I'm getting. Yeah.
- 3:44Well, you know, you got to think about these things a little bit.
- 3:46I have investors to keep happy, right.
- 3:48And then you know you have some news today.
- 3:50I mean it sort of starts the groundwork of the integration.
- 3:53Can you talk a little bit about how AI actually fits into data
- 3:57bricks as business? So this is, we basically have to
- 3:59now having spent six months together, figure out what is the
- 4:02strategy of the company, how do we combine the two companies?
- 4:04OK, so they're awesome at generative AI.
- 4:07We have a data platform. How do the two?
- 4:09Together and it turns out that basically there's something you
- 4:12can do that's pretty unique and I think this is what's gonna
- 4:14happen with all these data platforms.
- 4:16With data platforms, I mean data bricks, Snowflake, Bigquery, you
- 4:18name it, all these. In the future, I think what will
- 4:21happen is that you will basically leverage something
- 4:23like they had which we call is data intelligence where you can
- 4:26generate these generative AI models for each of your
- 4:30customers. And they really understand
- 4:32deeply the semantics of the data of the enterprise Oregon
- 4:36organization. So each each customer you have,
- 4:38the models that you create, understands exactly the jargon,
- 4:43the priorities, everything. So what that enables is that
- 4:47basically today in a large organization that uses a data
- 4:50platform, you need to have people who know how to code the
- 4:52right Python or SQL. But with this you can basically
- 4:56enable anyone who can speak English or any natural language
- 4:59to ask questions and you can get them the answers.
- 5:01So I think it changes everything.
- 5:02We call it Data Intelligence Platform.
- 5:04So you can in plain language, query data bricks against your
- 5:08data, right? Is that available now or is that
- 5:11it? Is, and I mean it's being
- 5:12improved continuously. So it's early days still, but I
- 5:15think the concept of making data interactive is really what.
- 5:19Brought this together and made it, made it happen.
- 5:21Because it's a It's a very natural thing.
- 5:23Like you want to customize, personalize that interaction.
- 5:26And you want to make it something where you can start
- 5:29driving business value across the company.
- 5:31Which means you can't just have people who know how to write SQL
- 5:33queries. And do you see actually
- 5:35providing foundation models as part of your business?
- 5:39Absolutely, Yeah. I mean, Foundation models I look
- 5:41at as a starting point for customization.
- 5:43I mean, sometimes it warrants building a whole new model from
- 5:46scratch depending on what the the types of data are.
- 5:48Like if it's. Very particular kinds of jargon
- 5:51or very particular kinds of knowledge that have to be
- 5:52embedded. This is.
- 5:53Which by the way, this is like that, that's your sport.
- 5:55That's what you guys did for Living Mosaic ML, building large
- 5:58language models from scratch. That's right.
- 6:00On custom data that enterprises had, right?
- 6:03Exactly, yeah. And you know, we want to make it
- 6:04easy. So where we see patterns, where
- 6:06different companies use things, we're going to build great
- 6:09starting points for that. There will always be cases where
- 6:11people need to do a higher level of customization.
- 6:14Are you trying to commoditize foundation models?
- 6:16Like how much do you think foundation models will be
- 6:19something where there's a lot of value to unlock or it's just
- 6:22sort of a starting point for for other businesses?
- 6:24It's a little bit of both. I mean it depends on what your
- 6:26use case is, right. I mean if you go and invest lots
- 6:29of money like Open AI or these other companies, they're they're
- 6:32going after like a very different kind of market than we
- 6:34are. I think enterprises need a lot
- 6:36of customization. It's very different than
- 6:37consumer and I I think the the, the cost to do so actually is
- 6:42commensurately high. Maybe the volume of inferences
- 6:44is also commensurately higher, but.
- 6:47It's it's a starting point, but it's ephemeral.
- 6:49Like every They have a time to live, six months, eight months,
- 6:52maybe a year. Your model is irrelevant because
- 6:54there's going to be better architecture, going to be better
- 6:56data, better data cleaning techniques, this kind of a
- 6:59thing. Who do you think are 1/2 right
- 7:00now and quality? I mean, TPT 4 is still the king,
- 7:04right? I mean, I don't think anyone's
- 7:06beat it yet. But The thing is, in the
- 7:07enterprise, I mean, this is what he's saying.
- 7:09You know, in the companies that you work, there are three letter
- 7:12acronyms that no one understands what they stand for, right?
- 7:15If you ask Chat TPD, it won't know what that is.
- 7:16You have specific data that's super confidential in your
- 7:19organization. You know Chat TPD doesn't know
- 7:21what that is. And by the way, when you ask
- 7:23questions from that AI, you want it to really be accurate.
- 7:26There's going to be a need for organizations to custom train at
- 7:28least or fine tune or modify is that rag?
- 7:31I mean I feel like everybody in the valley is talking about
- 7:33retrieval, augmented generation. Is that you think the key or you
- 7:36think it is fine tuning and having a sophisticated?
- 7:40It's a part of it. I mean all of these are ways to
- 7:42customize the outcome, right? Sometimes you RAG makes a lot of
- 7:45sense when you have permissions that change on particular data,
- 7:48when you have data that's updated continuously, that's a
- 7:50great use case for rag. We look at this on kind of a
- 7:52spectrum of techniques. It's not like there's one thing
- 7:55that's going to, you know kill everything else.
- 7:57They're all, they all have different power and and
- 8:00capabilities. And keep in mind for RAG, you
- 8:02also have a large language model and you're combining it with a
- 8:04vector search database. So it turns out, if you
- 8:06actually. Custom train the large language
- 8:09model to be really good. At RAG, you actually get even
- 8:12better results. So we're doing that as well.
- 8:15The accuracy issue, right, you sort of signaled that this is
- 8:18sort of just getting started. So you know some people are very
- 8:21reluctant to roll something out without sort of certainty.
- 8:24I mean you have like medical organizations in some cases
- 8:27using it. How do you think about how
- 8:29accurate the queries are, even against customer data?
- 8:32And how did you think about whether to wait for perfect
- 8:35accuracy versus let people sort of try it and experiment?
- 8:39Well, I think this is actually one of those clear bright line
- 8:42differences between enterprise use cases and consumer, right.
- 8:44If you're doing something where accuracy doesn't matter or you
- 8:46just you know it's a writing assistant or something like that
- 8:48you you can get away with a lot more stuff where as an
- 8:51enterprise we can't. But at the same time we need to
- 8:54get people used to the flow and thinking about it like this.
- 8:57I think having you know sort of techniques where you can, you
- 9:00can have some suggestions from the AI, but alongside what
- 9:03humans do for now. That's probably a good paradigm.
- 9:06As we get more and more accurate, you learn to trust the
- 9:08systems and then that can be something that can start taking
- 9:10over. But right now it's not really.
- 9:12I don't think you should turn loose one of these systems on
- 9:15mission critical outputs, you know.
- 9:18And the way so in the data intelligence platform what we've
- 9:20done is that yes, you can ask in English, it can find you the yes
- 9:24you can ask in English. And it like gives you the
- 9:26answer. But there is a box you can click
- 9:27on and it gives you the query that it actually.
- 9:29So you can go under the hood and you can have someone that if you
- 9:32want to be dead. Sure that this is correct.
- 9:35That person can audit and look at.
- 9:37OK, let me look at the query under the hood in Sequel, Yeah,
- 9:40this looks good. We can put it in the board deck.
- 9:42You know, this is our financial prediction for next year and
- 9:44we're not going to get fired at the first Cerebral Valley like
- 9:46it was like Open AI versus like the open source world.
- 9:49Or it's like can we sort of cobble together enough open
- 9:52source projects to to sort of fight against Open AI?
- 9:55On the one hand, you know, I think with Facebook and Llama,
- 9:58we've seen sort of strong offerings.
- 10:01On the other hand, GPD 4 still seems sort of invincible in
- 10:05terms of like being the smartest AI in the room.
- 10:08I'm curious, are you guys still all in on open source?
- 10:12I know it's key to your identity.
- 10:13You you would release Dolly like what's the state of contributing
- 10:17to open source and how it fits into the strategy of Databricks.
- 10:20100%, I mean, you know, it's we think it's super important that
- 10:24researchers around the world do open research.
- 10:27And we have these open models that we can understand because
- 10:29we don't really understand these things.
- 10:31I mean, we understand how we built them, but we don't
- 10:32understand why they exactly work, right?
- 10:34It's kind of like a little bit like the isn't that terrifying
- 10:36to you as Aceo? And it's like we do.
- 10:38It is terrifying. Yeah, So, but how do we
- 10:40understand it then? Is it to have two companies that
- 10:42have two secret models that they don't want to share anything
- 10:44about? Or do we want the researchers,
- 10:47all the all the sort of labs around the world, to spend time
- 10:50trying to understand what's going on and make progress
- 10:52towards understanding how these things work and how we can
- 10:54control them and how we can align them and all those kind of
- 10:56things? So we think open source is
- 10:58essential. And actually unfortunately I
- 11:00would say, I mean I don't know if you agree or not, but it
- 11:03seems there are talks now, there's, you know, in certain
- 11:06circles that maybe we should ban open source large language
- 11:09models altogether. There's discussions in many
- 11:11countries where this is kind of coming up, which I think it's
- 11:15horrible, it's, it's it's horrendous.
- 11:16It's going to, it's kind of put a stop to all innovation and
- 11:20it'll just kill off the whole ecosystem and it won't help us
- 11:23understand what these models do. So I think it's absolutely
- 11:27essential that we have an open ecosystem that continues to
- 11:29thrive. I mean, it's ironic because by
- 11:31closing off models, you're actually going to ensure the
- 11:35thing that you were trying to prevent, right?
- 11:37Because I think we all believe in the AI world that like, these
- 11:39models are weird. They're they can do potentially
- 11:42very damaging things. I don't know how that's going to
- 11:44manifest. I think the people who profess
- 11:45to know are kind of full of shit, to be honest with you.
- 11:49Because we really don't know. And that's OK.
- 11:50But the the way we're going to figure this out is through many
- 11:53minds, many people, researchers in academia and different
- 11:56companies, building solutions, putting those into the world,
- 11:59seeing how they work, and then figuring out how to make them
- 12:01better. We have to increase access, not
- 12:04limited. Keep in mind also all the big
- 12:06innovations that we're leveraging today that made these
- 12:08possible was done in open research.
- 12:10It was before the, you know, shutdown of all of this stuff.
- 12:12It was pre 2020 releasing the transformer paper, right?
- 12:15Yeah, public, public research. Just to follow up on one thing
- 12:20you said, do you think a country will ban open source?
- 12:22Like do you see that in the cards where some country is
- 12:25actually going to move and do that?
- 12:27I don't know. I hope not.
- 12:28There's serious talks. If I had to say from the
- 12:30information I have behind the scenes, it seems in some
- 12:34countries the the camp that's winning is the anti open source
- 12:37camp right now because you know you have the biggest companies.
- 12:40Saying, hey, this is super dangerous.
- 12:43I'm creating it and what I'm creating is super dangerous.
- 12:45So please regulate me. And then, you know, the
- 12:47regulators are like, OK, they're telling me to regulate them.
- 12:49And, you know, and media is writing about how, you know,
- 12:52dangerous this could be. So they're freaking out as well.
- 12:55The public's freaking out. So, you know, it's kind of
- 12:57pointing in that direction. So right now, I actually think
- 13:00that it's going in the wrong way.
- 13:01I hope we can stop it. Because just trusting two
- 13:04companies to figure this out or four companies to figure this
- 13:07out, I don't think it's the right way.
- 13:08Yeah, you you guys want to be team open source, you're a big
- 13:11company sort of respected. How do you, how did that fit
- 13:14into your thinking on the Biden executive order?
- 13:17Did you guys stake out a position?
- 13:19What is the position? Did you think that was sort of a
- 13:22good middle ground or how do you think about the executive?
- 13:24Order this This is just the first sort of step, right?
- 13:27For the writing, the executive order.
- 13:28It has some limits and so on. It didn't actually weigh in on
- 13:32whether open source is going to be allowed or not.
- 13:33It kind of mentions it making weights free.
- 13:36So we hope that if this becomes law or the continuation of this
- 13:42absolutely is not going to sort of ban or put a stop to open
- 13:45source because that's going to be essential to figure it out.
- 13:47And also by the way, there's worries about what about other
- 13:50countries, they could just pick up this open source stuff.
- 13:52It'd be like basically giving away our IP to other countries.
- 13:55But what would you rather do that all countries in the world
- 13:58are leveraging your technology stack or that they're building
- 14:01their own? Proprietary thing that they
- 14:03have, right. So, so we hope, we hope that
- 14:06it's going to continue to support open open source, open
- 14:08research, but we don't really know, yeah.
- 14:11And I think for the executive order, there were some some good
- 14:13things. I think NIST is a good natural
- 14:15home for some of this stuff, which I thought was a good, good
- 14:18thing. Focus on the organization that's
- 14:19going to oversee it. That's.
- 14:20Right under under the commerce branch and I think that that
- 14:24made a lot of sense. I think focus on transparency,
- 14:26maybe some data tools or on lineages kind of stuff.
- 14:29There are opportunities here in terms of market.
- 14:31Which is which is a good thing. The the places where I think it
- 14:34might be kind of not dangerous but like not super relevant or
- 14:38when you start putting compute limits in because these things
- 14:40change constantly. Something that was really big a
- 14:42year ago is really not that big anymore.
- 14:45So I think it was 1 E 26 FLOPS. I mean OK what precision what
- 14:48you know there's so many different things that how that
- 14:50can be interpreted. So I I don't think it's a great
- 14:52idea to start dictating these kind of limits because we just
- 14:55so are you do you oppose it on? That well, I I'm not.
- 14:58I'm opposed to that part of it. I don't think it dictated in a
- 15:00hard way. It was kind of soft.
- 15:02I guess for me it's still an open question.
- 15:04How quickly can those limits be changed?
- 15:06Are they going to have to go through a whole government
- 15:08committee or something like that and take a year to change or is
- 15:10it something that's that can be highly variable?
- 15:13And then the limits are pretty high right now for this year or
- 15:15next year. But yeah, in five years you look
- 15:16back and it's, you know, it's silly, probably.
- 15:19Are you worried about existential risk or is that just
- 15:22a way to hype up the industry? I mean, Dario at Anthropic gave
- 15:24an interview where he said he thought there was like 20%
- 15:27chance something like really bad happens and he runs, you know,
- 15:30one of the top foundation model companies.
- 15:32Do you feel that way or how much risk do you see?
- 15:35This is what I'm saying. It's like, hey, I'm going to
- 15:37build a huge model. And by the way, it's going to
- 15:39take over the world, you know, regulate me.
- 15:41What do you think of that? And I.
- 15:43Think that 20% number is, But we're not exactly on the same
- 15:46page. I think with existential risk.
- 15:48I love disagreement. Let's go.
- 15:49That's OK. Are you with your boss right
- 15:50now? I think you're, I mean, I don't
- 15:53want to put words in your mouth. You're more in the camp of like,
- 15:55you know it's way less. And I'm a little bit like, you
- 15:58know, maybe it's a little bit more than, you know.
- 15:59But we're both actually on the side of, let me let me kind of
- 16:03like argue sort of his side. I do think that we're protected
- 16:06right now because these things can't reproduce themselves.
- 16:09We're simply protected by the following thing.
- 16:11It costs a lot of money and it takes a lot of time to train GPT
- 16:164. OK.
- 16:17And the scaling laws say if you want to have even better model,
- 16:20you better spend even more money and even more time, right?
- 16:24Just to get more intelligence. OK, so therefore I'm protected.
- 16:28These things are not going to reproduce themselves.
- 16:30Now if it was the case that you could for one cent cost too
- 16:34much. But if it costs one cent, and if
- 16:37it took one second to produce GPT 4, you can see, then you
- 16:40would start making a loop for loop, run a genetic algorithm,
- 16:44improve itself, and you could see how you could sort of get to
- 16:46this autonomous reproduction, but without reproduction.
- 16:49You know, I think we're we're safe, actually.
- 16:51We should still do research and understand this stuff.
- 16:53But I don't think the essential risk is like immediate or that
- 16:55we should stop all research or stop all activity that's going
- 17:00on. It's it's sort of exaggerated
- 17:01and I think that 2020% number is sort of just completely random.
- 17:04I don't know what you. Think yeah, I think that 20% is
- 17:06way, way overcalling it. What we all kind of agree on is
- 17:08that there is some eventuality where AIS will become as
- 17:12intelligent, if not more intelligent than humans.
- 17:13I don't. I don't think anyone will really
- 17:14argue with that. It's more the time scale.
- 17:16And I look at it like, OK, hang on, If you start putting these
- 17:19restrictions on now, you're actually going to make it so
- 17:21that fewer people are working on this problem.
- 17:23That's a bad thing. That's what I'm worried about.
- 17:25I think at some point that will be relevant, especially when it
- 17:28becomes one cent to do a G PT4 style model.
- 17:31Also, a lot of time, a lot of the things I've seen have been
- 17:34rhetoric that's saying like, oh, I can see a model escaping and
- 17:36this and that. Really, you can look at that in
- 17:39in terms of what is true today. Computer viruses are self
- 17:42replicating. They can actually escape.
- 17:44They can do all of these things. Is this really a new problem or
- 17:47you just slapping AI on it and making people scared about it?
- 17:49That's a problem I have. So let's let's look at the
- 17:52things that are real, real threats that we have today and I
- 17:55think this information might be one of them.
- 17:56Let's focus on those, right. Let's focus on robotic safety.
- 17:59Like, I mean there was AAAI think a factory worker that was
- 18:02killed by a automated robot just two, two weeks ago.
- 18:06So these are real threats in the in the short term, what's going
- 18:09to happen in 20 years? I don't know.
- 18:11Let's. Or the OR the use cases, you
- 18:12know, for putting these AIS into weaponry.
- 18:15Absolutely. Maybe we should look at that.
- 18:16Maybe we should regulate that, right?
- 18:18Maybe that we shouldn't go crazy with that.
- 18:19All right, I want to try and thread a sort of nuanced
- 18:22question. I mean, on the one hand, being
- 18:25more of an AI company seems great for the Databricks story
- 18:28on a March towards going public someday.
- 18:30I feel like there's a huge appetite for like an AI company.
- 18:34On the other hand, it's really expensive to run.
- 18:37Like how did you think about the trade off in terms of actually
- 18:40deploying stuff when it's very costly?
- 18:42And how did that fit into your calculus of maybe moving towards
- 18:45an IPO? And can you give us any sort of
- 18:47update on where that stands? Yeah, the IPO plans got smashed
- 18:51with the acquisition. He He destroyed the.
- 18:56He destroyed. That destroying the PL.
- 18:58everyday, Did it delay? Did it delay when you make a
- 19:00bubble? No, it it doesn't actually we we
- 19:02run, we're very careful about this stuff.
- 19:03You know it's all part of the plans and you know the way I
- 19:06would think about it is we just have a much bigger budget to
- 19:08absorb these things. So you know it's different for a
- 19:11start up with 5-10 people that doesn't have any revenue yet to
- 19:14spend $100 million on GPU's. You know our budget, annual
- 19:17budget with or without these things was in the billions
- 19:20anyway. So you can absorb these things
- 19:22plus. These guys did a really good job
- 19:25of, I mean, their revenue was growing really fast.
- 19:26So they're also selling the GPUs.
- 19:28So we're cooking it doesn't micro public next year.
- 19:32That's a great question. We are watching the public
- 19:34markets. We looked at the recent IPOs
- 19:37that happened. Actually they haven't been smash
- 19:40hit. It's kind of wobbly.
- 19:42So you know when the time is right and the markets are open,
- 19:44we will go. It's not something that we
- 19:46obsess over right now to be honest.
- 19:47We're just, you know, there's so much demand for AI.
- 19:49We just want to satisfy that and continue innovating.
- 19:52Do you have a Microsoft partnership right now?
- 19:53We do. And what's the relationship?
- 19:55That's great. Thanks for asking.
- 19:57All right. You got, you know, I'm a
- 19:59diligent reporter at the end of the day.
- 20:01OK, last give it just real quick because we're out of time.
- 20:04But give us a piece of advice for people here who want to make
- 20:07like a deal like you guys did. What is?
- 20:09What is the advice you give in terms of building the
- 20:11relationship or doing something like?
- 20:13That come to this event and go to the party afterwards and grab
- 20:15come. To the party and sell your
- 20:17company for a billion dollars, right?
- 20:18That's what happens. And answer the phone calls on
- 20:20the weekends. You know I I think these these
- 20:22events is it's it's it's serendipitous right.
- 20:25There's no good way to really architect this stuff to happen
- 20:28but I think relationships matter.
- 20:29I think FaceTime getting to to meet people and know them
- 20:33actually really matters and that's how we build trust.
- 20:35I mean, honestly the reason this happened is that when I
- 20:38interacted with Ally and the other Co founders of of
- 20:40Databricks and they interact with us, I was like, we're not
- 20:42going to agree on everything and that's OK, but we're going to
- 20:45figure the shit out. Whatever it is we're going to,
- 20:47I'm I'm confident we can figure it out.
- 20:49That's what matters. Can I give some some insight?
- 20:51Scoop, I think, like, build an awesome company, right?
- 20:53You know, when we looked at Mosaic, we were like, wow, they
- 20:55get really good people. You know, they're sort of we're
- 20:58the same DNA, So that was important.
- 21:00So hire phenomenal people. That matters.
- 21:02Second, they had an awesome business, you know they were
- 21:04growing like crazy. You were like 1,000,000 ARR in
- 21:06January and there was 20 million ARR by the time the acquisition
- 21:09May or June. We were talking about April, May
- 21:12time frame. So you know the business was
- 21:13growing. So you had actually a good
- 21:14business model foundation to it as well.
- 21:17So I think you know, have a good business model and have great
- 21:19people, then you know, people will be very, very interested.
- 21:23Ollie Naveen, thank you so much for all your support and coming
- 21:25here on stage. All right.
- 21:27Next panel. Thank you.
- 21:31That's our episode. Thanks so much for listening.
- 21:33Shout out to Max Child and James Wilsterman, my Cerebral Valley
- 21:37AI Summit Co host. Thank you to Riley Kinsella, my
- 21:40chief of staff Gabby Caliendo at Volley, who's been instrumental
- 21:44on putting the conference together.
- 21:46Thanks to Young Chomsky for the theme music.
- 21:48Please like, comment, subscribe on YouTube, give us a review on
- 21:51Apple Podcasts, and please subscribe to the sub stack
- 21:55newcomer.co. Thank you so much.
- 21:59Goodbye. Goodbye.
- 22:00Goodbye. Goodbye.
- 22:02Goodbye, Goodbye, Goodbye.