theCUBEd Awards | Daz Williams, InFlux Technologies (AI Luminaries)

Clips
News
More from theCUBEd Awards

Daz Williams

Chief AI Officer

Influx Technologies

How InFlux is reshaping AI with decentralization to cut costs and boost efficiency

The vision of Web 3.0, the decentralized internet, now encompasses artificial intelligence.Starting with a blockchain system, InFlux Technologies Ltd. has grown into one of the most promising AI firms, according to according to Daz Williams (pictured), chief AI officer of InFlux. The company has created a decentralized marketplace that gives developers a more efficient, economical way to access enterprise-grade AI.“Developing AI products, it’s building the airplane while it’s still flying and being the first, and discovering problems that no one else has had before,” Williams said

play_circle_outline Celebrating winners of Tech Innovation CUBEd Awards

play_circle_outline The Significance of AI in Today's Technology and the Evolution of Decentralized Architectures

play_circle_outline Bringing Web3 era with blockchain and decentralized applications

play_circle_outline Making decentralized app development easy for developers

Info
Transcript

Daz Williams, InFlux Technologies (AI Luminaries)

Daz Williams

Chief AI Officer Influx Technologies

In this insightful episode, Daz Williams, Chief AI Officer at InFlux Technologies, joins theCUBE to discuss the dynamic landscape of Artificial Intelligence (AI). With his profound expertise in developing AI products, Williams is awarded for his pioneering work in AI at the inaugural Tech Innovation CUBEd Awards. Conducted by theCUBE Research, the conversation delves into the evolution of AI technology, highlighting Williams' contribution to distributed computing as well as his vision for the future of AI-driven applications.

According to Williams, the... Read more

explore Keep Exploring

What event are we here to celebrate, and who is the winner of the AI luminaries category at the Tech Innovation CUBEd Awards? add

What is your vision on the innovation of AI right now from your perspective? add

What core problem did the person innovate on when bringing in the era of Web3, including blockchain, decentralized applications, D'CENT app, NFTs, and decentralized AI? add

What is the process for getting started with building a decentralized app on the RunOnFlux website? add

bolt Powered by CUBE AI

Daz Williams, InFlux Technologies (AI Luminaries)

search

>> Hello, welcome back to theCUBE, here in our Palo Alto Studios. I'm John Furrier, host of theCUBE. We are here to celebrate theCUBE's Tech Innovation CUBEd Awards, our inaugural event. We're celebrating the winners. The competition was fierce. We received a ton of exceptional entries in multiple categories. One of my favorite was the AI luminaries category, and we've got a great guest winner here, Daz Williams, Chief AI Officer at InFlux Technologies, AI luminary. Congratulations. You're in the first inaugural category of AI luminaries in this program. We're going to have another one this year. Congratulations. Welcome.

Daz Williams

>> Thanks, John. Yeah, it's a great privilege, of course. And to be the first, and that's kind of what it's felt like all along in this journey on developing AI products. It's building the airplane while it's still flying and being the first, and discovering problems that no one else has had before. So yeah, it's very fitting.

>> It's great. I mean, you're in the inaugural class of an AI luminary. I'm sure this is going to be a very hot category. It's going to continue for generations. I love the AI wave right now, because one, I love software. I love distributed computing. I think we're in a systems revolution. People who listen to our CUBE pod with me and Dave, and hear my rants on theCUBE know that cloud computing was awesome. We saw that wave come in since the 15 years of theCUBE. But now as it goes next level, you have full distributed computing paradigms, you have new decentralized architectures, a true generational shift, I would say multiple decades since we've seen anything like this is happening. Gen AI is certainly powering the hype and also some of the ways people are writing software for sure, but there's still the middle and the infrastructure going on that's still a lot of advancement. So when that starts to get taken care of and it is, there will be a tsunami of AI applications, some of them low code, no code, code assistants, to humans driving more creative innovation, more ways to do things. Daz, tell me your vision on the innovation of AI right now from your perspective.

Daz Williams

>> I mean, GPUs, right? Right now it's the currency. And if you saw the launch of the 5090 with NVIDIA, it was just a sellout. I was one of the first to arrive in line at MicroCenter in Dallas, even though I got there at 4:00 AM, I was right at the back, so kind of sucked that they only had nine available. But that just shows the demand right now. And that's what got me interested in InFlux, where I'm currently at. And it's being able to bring distributed computing globally, by allowing people to bring their own GPUs to the network. They get rewarded through rentals, and then we are able to harness this massive amount of processing power, which would either be far too expensive for a company like ours, or just unavailable. And so, we are able to throw our product FluxEdge, bring all of these GPUs together, and I was able to create a software that allowed us to deploy all of the AI models, and bring them to every developer, and allow them to access this huge amount of process and power. But very affordably.

>> Not to date myself Daz, but I remember the term grid computing. If you remember back in the day, what you guys are getting at is again, a new way to leverage efficiency through kind of a decentralized way. Talk about how important this is and scope the opportunity from where we are today. Because again, not to go back and talk about some of the news cycle around GPUs, but you talk about people in the trenches, they're all whispering, it's a public secret. There's all these cycles that are wasted, sitting there idle. There's all kinds of opportunities around leveraging a lot of different piece parts. Not everyone can put gigawatts of power in their home. So talk about the evolution here and why this is important. Scope the magnitude of the opportunity, kind of where we are right now and where it will go.

Daz Williams

>> Yeah. So you've got different ends of the market, from the corporations and the big giant data centers. And then, what we are trying to do is, we're trying to address the other end of the market, where it's unused GPU power. And most people's homes out there that's got these with 3090s, 4090s, now the 5090s, we can be utilizing this enormous amount of power whilst people are asleep. They can contribute their GPU to our network and we can have that globally. And what that allows us to do with our AI technology is then have that globally distributed in a load balanced way, so it can provide anyone in the world super fast AI services. And it wouldn't be possible without the average little Joe at home being able to plug his GPU in. It becomes, it serves one end of the market, which allows us to have this power that we wouldn't otherwise have.

>> Talk about looking at the notes here from the Web3, your website. So bringing in the era of Web3, that meant a lot about blockchain, it kind of scales up to decentralized applications. What's the core problem that you innovated on that you could point to? Because we saw that D'CENT app kind of wave coming and then NFTs came in, and now you're starting to see decentralized using AI. So there's a lot of AI in decentralized, people think it's just gen AI for all these proprietary models and small language models, and these multimodal models, but there's a lot of AI going on in here too. So talk about what's going on there. Can you share your thoughts on this?

Daz Williams

>> Yeah, so you've got the giant AI models, like the likes of GPT4, hopefully now they're coming up the GPT5 model. Then you've got Claude and all of these other smaller models. And what about the average developer out there, who wants to harness this type of power but can't? And so, what we are able to do is create a software that allows them to put this onto their own computers and run these large language models from GPUs on their own machines. It really does transform the whole industry upside down and how this works. And InFlux was able to start off with Bitcoin, with mining, with the mining and the Flux on the blockchain, and realized there was so much unused power that was just being used for mining, which we could then utilize and convert over to running the large language models, which is what we're able to do.

>> Daz, you're talking about when you started coding, how'd you get here?

Daz Williams

>> So go back 2016, I created the first product called Koddi, and it was specifically designed for eCommerce sites. And the idea behind it was very, I mean, we were very early on in our thinking, but it was how can you take tens of thousands of products on a website and narrow them down through natural language, through conversation? And we were able to bring this together and do a great MVP, and you would essentially say, "Hey," if it was a fashion website, "I'm going to an event at the weekend and I'm looking for something formal to wear." And so, it would understand the context behind what you're saying and then present on the screen, real time, all the different products that would match that search. And we were way ahead of our time, 2016. I mean, it didn't feel that long ago, but in AI terms it was a long time ago and I think people just weren't ready to take that leap yet. But fast-forward to now, and we're solving similar problems. So if you imagine companies that have hundreds of thousands of documents and they want to build to provide knowledge bases, whether it's for their staff or whether it's for their customers, now how are you going to find the perfect product, the perfect document from them hundreds of thousands of different documents? And what we're able to do is not only introduce a RAG system, which is the new buzzword these days, but it essentially what allows people to search through lots of documents very quickly. But we've noticed, when you do this, it's a very poor user experience. The documents I get returned are great, but how does it solve a real problem? How does it solve that user's problem? And what we're able to do is now bring in a conversation for the LLM, to then define and get down to the exact document they need. So let me give you an example. If it's a mechanic, and actually let me give you a better example, John. The example is in a healthcare industry with doctors who are trying to do an MRI scan. And so, all of a sudden now they're having problems with the machines, so they're now need to find help. And what machine is it? If there's 10 of these different machines that's been built over the last 10 years, we'll have all these documents. But how do we know exactly which document it is? And so, the first part of this is to make sure all documents, they have rich metadata, and that's part of the RAG. But then the next part is a natural conversation with the LLM to narrow down the results. Have you ever come across that program, ?

>> No.

Daz Williams

>> It's a game and it basically says, "Think of any character alive or dead." And you basically can choose any character you want and asks a series of questions, and all it's doing is it's halving the results every time it asks a question, until it eventually arrives at the perfect one. And that's what we've managed to achieve with what we're building, is to take these thousands of documents and narrow it down to the one which will allow the LLM to more accurately answer the user, so it becomes more of a conversational flow.

>> And manage all the data, so you basically had the efficiency with all the data. Is that what Flux AI is doing or is that separate software, or is that one and the same?

Daz Williams

>> Yeah, that's part of what Flux AI is building. This is an enterprise product.

>> Got it.

Daz Williams

>> And we had a customer who reached out to us and said, "Look, I've got all these products, but the doctors just can't find what they're looking for easily." And so, when we used their current knowledge base, it uses some kind of ML technology, but the results are great, but how does that help the doctor? They have to go through every result. What we want to do is get it down to that single result, and that's through enriched metadata and through conductive reasoning.

>> So how would you talk about that innovation? Was it the scale? Was it AI driven? Was it data driven? What was the innovation? Unpack how you got to that search solution from... And what, if any of the knowledge graph made it inefficient or inadequate?

Daz Williams

>> All right, so if you right now was to go and look and try any RAG service out there, and all it'll do is it'll present you with a list of results. And if we're talking about MRI machines and the hardware, and the hospitals, there might be five different manuals for five different machines, but the doctor wants the one for one particular machine. So naturally, what would happen in the human world, and that's what we're trying to mimic, and that's how we got here, was by asking a series of questions. And so, the first question you would ask is, "Well, what year is it? Or what's the model of the machine?" And that data is all there, but the current RAG or knowledge graph systems out there that just don't do it in this way, where it's able to have a natural conversation to then help you arrive at the exact product you need. And like I said, it's using that, halving it every time results until you get to the one you need.

>> Well Daz, it's great to have you on as a luminary. I've got to ask you the question. I'm totally into dApps. I want to start building, what do I do? How do I build dApps? Am I using Ethereum? Is there other... Is competition now with Ethereum, what do you guys have? Are you guys going to power the software? So in all hardware, you got to run software on it. What do you see as the dominant dApp framework, approach? If someone wants to jump in, maybe doing some SaaS coding, I want to jump in and build on a decentralized infrastructure. What would you say to that?

Daz Williams

>> So building a decentralized app, it's so complicated. There's so many different choices out there. And what we're able to do, it's InFlux, on our RunOnFlux website is, we made it so easy to get started. You can be a developer, but with having no knowledge of the blockchain, you can log right on there and it'll take you through a series of questions, and it'll help you choose the right solution and then allow you to get started right away. So I was really pleased to be part of that project with a wonderful team. One of the first things I got involved with at InFlux, and we ended up with such a great product. And then the world moves fast, and then we got in all the AI different products come out and that's where we are now. It's just such an exciting journey.

>> All right, so what's the coolest thing you're working on right now?

Daz Williams

>> Yeah, it's got to be, we are calling it's like D3DS, which is the Document Distillation and Deductive Reasoning. And that allows us to document hundred thousands of documents and allow us to have a natural conversation to get to the exact one we need.

>> How do you get that performance? What's the secret sauce?

Daz Williams

>> No, I mean the secret sauce is just knowing all these different components and there's a lot of great people out there that's done a lot of innovation, and they've actually invented these wonderful products. But being able to bring all of the different pieces together, even in a simple way, and allow that to become a unique product in itself. And that's innovation right there, but also built on top of other really innovative products.

>> Yeah, I love that you bring up the search piece, because I always said search has always been the killer app. Find what you're looking for. That's not search, that's just a value statement. And that's extending into large scale data sets over a decentralized architecture, centralized, decentralized, all kind of abstracted away for the user. That is what gets people excited. Where does this go next? Where do you see this going? What's the next dot that you're going to connect? Take us out a few dots, what happens next?

Daz Williams

>> Yeah, I mean, if you go back to the early days of web design, web development, web hosting, it was essentially you had to be an expert in lots of different areas, DNS, domain systems, coding, setting up the host, all these different things. And then, of course you got wonderful innovations from some of the big companies. AWS, I just made it so easy for someone to just do everything all at once. So you didn't necessarily have to be an expert in all these different products, the GoDaddys of the world, all these different products out there that just made it so easy and simple. That's what we're experiencing right now with AI, is you've got to be an expert in all these different areas. The LLMs, architecture, what GPUs do I need, all these different things. And so, I think right now the solution is to make it so easy for anyone.

>> Daz, it's great to have you on again, Chief AI Officer at Inflex Technologies, winning CUBEd. You've been CUBEd on AI Luminaries. Lot of great action with you guys, what you're doing, what your company's doing. Obviously I was just talking about this last week in theCUBE pod. There's a huge liquidity coming in the entire crypto infrastructure and application space. The shift is back, people coming back into the US, the world's kind of uniting. I think it's going to be a great next five years for this next generation infrastructure. And again, the work you guys are doing over there is great. Thanks for coming on, and congratulations for being an AI Luminary in theCUBE Awards. You've been CUBEd.

Daz Williams

>> Thanks very much, John. Appreciate it.

>> Cheers. Okay, we're going to continue theCUBE Awards. We're going to have another entry coming in. Again, we did this as our inaugural event, huge response, 43 judges and more coming in. Hopefully it'll be a community award, maybe get in the blockchain, build an app for this and get Inflex Technologies to help us out. We had to completely democratize this, but we'd love what we're doing. And again, thanks for watching theCUBE Awards. I'm John Furrier, your host. Thanks for watching.