The Exchanges

Q And very clear. So is that, is that the beyond Victor? Is that the future of how LLMs get deployed in the world where the quality of answer matters?

A I, I think there is two ways to have, uh, GPT for your own data. The first way is to do what's called model fine tuning. That is here, Matt, go read all of these 10 books and relearn all of these concepts in your neural network. Uh, I think that way is susceptible to hallucination. That way is very expensive because retraining a neural network actually is very costly to retrain it. Uh, that way is very slow. A new fact coming in, a new book coming in, you have to retrain. It's not going to be available in the outputs until weeks later. Uh, the, the grounded, uh, generation approach is real time. A new fact comes in is showing up in the answers right away. There is no hallucination. I mean, you minimize it significantly and the cost is a hundred times cheaper. So I am biased because that's what I'm selling. But yes, I think, I think that is the new way. The new way is going to be the combination of excellent retrieval models that know how to fetch the facts and then domain specific models, whether that be legal or finance or health that know how to take these facts. And convert them into a response, uh, for the question the user is asking.

AI assessment note: “So I am biased because that's what I'm selling. But yes, I think”

Q a little bit of a, you know, lifetime dream that just came true. Um, I'd, I'd love, uh, maybe, uh, last question from me before, uh, giving a chance to people to ask. Question, uh, again, like zooming back out. What do you make of, uh, the whole pause idea in research, uh, stopping all, uh, you know, AI research for six months? Does that make any sense to you?

A Uh, so first show of hands, how many of you heard about this letter asking for a pause in research for six months? So 50% of the room, uh, raised their hand. So this letter came out a few weeks ago, and a number of people signed it, including Elon Musk. Uh, I, I think it's mainly Elon Musk wanting to catch up. So he's asking, uh, OpenAI and everybody else, just pause until I buy all of my GPUs and catch up with you. Uh, I, I'm half joking there. I, I actually partially agree with the spirit of the letter. I, I don't think we should pause anything. By definition, whenever you have economical efficiencies, like the industrial revolution, or the mobile revolution, or now that this amazing gen AI revolution, we're not gonna stop. We're gonna make it happen. We're gonna make it work. We're gonna figure it out. We, we know that's how uh, capitalism works. Uh, however, we, the caution of being careful how these technologies can be abused is definitely something I'm very, very concerned about. So today, there is cloning technologies that can clone our voice very easily. I can, I can clone your voice right now. Call up your wife and tell your wife, I'm really stranded in this corner by Broadway and whatever, and she will show up thinking it's you. And there's nothing that prevents me from doing that today. What these, uh, folks have a responsibility to do is to make sure there is a very…

AI assessment note: “I don't think we should pause anything. By definition, whenever you have economical efficiencies”

Q um, for people to ask questions. But let's talk about Cloudera a little bit. So you built this incredible company. So you're a co-founder of it. You guys started in 2008. The company went public. Still is public. Still is public. Any thoughts about sort of where we are in that journey? So Cloudera was one of the key distributions of Hadoop. Where are we on that? On that journey.

A I mean, Cloudera still is the key distribution of Hadoop, so I should just very quickly say I was with Cloudera for 11.5 years since inception until the point when I left. When I left a few months ago, Cloudera's annual run rate in terms of revenue was about seven hundred million dollars per year. That's not shabby. Many companies dream of achieving that goal and continues to grow at a healthy rate, and the market cap is around three billion dollars. Uh, so overall, by all measures, Cloudera is a great success. Uh, I believe that Cloudera could be a lot bigger and a lot more successful, uh, but the market has been penalizing Cloudera, rightly so, for, for not, um, capturing the cloud's, uh, components as, uh, sooner. And, uh, a couple of very good counter examples to that is what MongoDB did. So MongoDB started betting on the cloud A lot sooner than Cloudera did, and hence they were able to share the revenue growth because it was sizable within the cloud. Cloudera is still at a point where we can't share that because it's still a smaller function compared to what we have on-premises. Our on-premises business, I shouldn't say our, their on-premises business, because I'm no longer there, has been growing in a very healthy way, but investors at large are worried that that's a mainframe business, right? That on-premise business is a mainframe. Now, I don't agree with that because o…

AI assessment note: “overall, by all measures, Cloudera is a great success”

Q Um, so, uh, if I was to deploy Victara in my enterprise, I would basically connect to lots of different, uh, sources of data, and, um, in the traditional world of search, you would go and index the content. Is this completely different, or is there some overlap between the way traditional search work and neural, and the way neural search works?

A Very, very, very good question. So, the previous wave The previous wave or the classical way for how we did search all of us and the way we have been trained on it with Google and many other search systems that we have been using is what's called keyword search, right? That's why we're trying to match the keyword you're looking for with what's in the content. So you search for weather balloon and we try to exactly match weather balloon. We don't try to match the meaning of weather balloon. The meaning of weather balloon should also match UFO. Uh, now, nowadays it does. It wasn't before, but now it does. So, so, uh, but an intern that has a smart brain that understands the meaning would do that matching. So these newer techniques for doing information retrieval, uh, they are based on, again, a neural network model that is almost as smart as us in terms of mapping from language space to a meaning space. And the beauty of these models is they can do it across languages. So whether that is weather balloon in, in English or Chinese or Japanese or Korean, all of them in the human language space, the neural network maps it to the same meaning in the meaning space. And that now allows us to, uh, always fetch the proper meaning for what you're after. Uh, what we provide at Victoria is an API that allows you to do that. So we sell to developers. We have an API that allows you to upload a…

AI assessment note: “the classical way for how we did search... is what's called keyword search”

Q Um, maybe give us a bit more color. Like what does that mean exactly? If you're an open source vendor, why should you work with GCP, uh, in priority and what's the actual, uh, sort of implementation of all of this?

A Yeah, I mean, our view is to establish the standards, and that's actually not just the Google view, like, that's also my personal view, is the power of open sources establishing standards. So, but it's not the only way of doing things. Like, again, if you go back to the, ah, analogy I gave with Android and iOS, Android is open source, iOS is closed source, iOS is very successful. And I think the same thing can happen in the enterprise as well. So you can have a closed source approach, which, You might claim that the Microsoft Azure Arc approach, or again, I keep mentioning the names, or the AWS Outpost approach, uh, can also be successful in achieving this vision I, I, I laid out here. That the, the difference is, I think when you're trying to win in the enterprise, history has shown that open source, when it comes to standardization at the platform side, wins. And the big, the best historical example of that is, uh, Linux, obviously, right? So before Linux, we had HPUX, we had IBM AIX, we had Solaris, we had SunOS, and many, many others. And when Linux came in, it won over everything else, right? The others exist, but in small fractions compared to what Linux became. So we have that same belief as well. Like we think to succeed in the enterprise on premise, in the cloud, I don't think it makes a big difference, by the way, if it's open source or not, to be honest, because you'…

AI assessment note: “our view is to establish the standards, and that's actually not just the Google view”

Q Great. So, um, back to, um, Victara as a business, so, um, how do you sell, who do you sell to, uh, how does the, yeah, use cases, how does the business side of the story work?

A Yeah, excellent question. So, I, I don't know if Will is still here, because Will was talking about open source, I have went through open source and I have the bottle scars from open source and how brutal Amazon is at competing with open source. Uh, open core does not work against Amazon because open core, you're building the core open and then manageability, security, reliability. That's your pro pro Amazon is really good at that shit. Like that's, that's the, they know how to do security. So, so it becomes very hard to compete with them. So I can give him some tips and lessons there. If you look at the most successful, uh, software company, That IPO'd in the last 10 years, who would come to your mind as that? The company that IPO'd that was most successful valuation wise?

AI assessment note: “Will was talking about open source, I have went through open source”