Cloud AWS re:Invent Coverage | Alan Chhabra, MongoDB

Clips
News
More from Cloud AWS re:Invent Coverage

Alan Chhabra

EVP, Partners

MongoDB

Blueprinting the future: MongoDB and AWS partner for AI database integration

As artificial intelligence reshapes industries, AI database integration has become a cornerstone for connecting advanced AI capabilities with practica...

play_circle_outline Evolution of stack to provide value to developers

play_circle_outline Enhancing Developer Skills Through AI Integration for Inference: A Training Model Approach

play_circle_outline MongoDB AI Applications Program (MAAP)

Info
Transcript

Alan Chhabra, MongoDB

Alan Chhabra

EVP, Partners MongoDB

John Furrier hosts an analyst session with Dave Vellante. MongoDB's EVP of partners, Alan Chhabra, discusses the evolving stack providing value to developers. The focus is on the intersection of AI and database technology, with implications for application development and ROI. MongoDB announces the MongoDB AI Applications Program with various partners to accelerate the transition from experimentation to production. AWS re:Invent offers a platform to connect with industry peers and discuss challenges and opportunities in the AI landscape.

explore Keep Exploring

What is the significance of the partnership between MongoDB and Amazon Web Services (AWS)? add

What is the significance of combining the world of databases with vector components and popular language models in order to bring agentic workflows or RAG workflows into applications for companies to make money? add

What is the MongoDB AI Applications Program (MAAP) and what does it provide customers with? add

bolt Powered by CUBE AI

Alan Chhabra, MongoDB

search

>> Okay, welcome back everyone to the CUBE coverage. I'm John Furrier, your host of the CUBE with the whole team here. Dave Vellante is in the analyst session. Stay zero as we start getting ready for the big day one tomorrow, Tuesday, the big news. Of course, a lot of announcements, a lot of talk about January, but also the stack is evolving to provide great value to developers. We've been covering that on SiliconANGLE, of course, in theCUBE. Alan Chhabra's here, the EVP of partners at MongoDB, company we've been covering deeply as well and is a love by developers. So when I think of developers in MongoDB, you guys go hand in hand. Everyone kind of knows who you guys do. Great to have you back on the CUBE.

Alan Chhabra

>> Yeah, thanks John. And by the way, it's a pleasure to be here at AWS. I can't even count how many the events I've been to. This one's probably the most special to me because our partnership's never been better.

>> Yeah, I mean, it's fun to watch, especially Amazon coming back from the, I would say negative press people, "Oh, they're behind on AI," which they really weren't. But they've been so focused for the past, I think year and a half, you've seen what they've done. But as AI starts to emerge, you start to see this formation, the fog's lifting. At the end of the day, it's going to be developers. And Matt Garman told me on my exclusive preview, they're going to get back to the basics. What that means is build value with developers, and there's developers inside companies, there's startup developers. They got to run on a database. Now you got inference. So I think the world is spinning to this MongoDB value proposition that you guys have been on for multiple years, which is, hey, how do I make the life of developer easier and more productive and scalable? I mean, I'm oversimplifying, but that's pretty much what you guys have been saying, right?

Alan Chhabra

>> Yeah, yeah. I mean, I give the hyperscalers a lot of credit. They're giving developers a platform to get access to tools, whether for AI or for specialty applications. The hyperscalers have always been there for developers. And when MongoDB started, we started the company from developers. Our founders built their own modern database because they wanted to make it easier to use data, whether it was for metadata use cases, operational, and today, vector data use cases. So we have hundreds of thousands of applications using MongoDB around the world, 50,000 paying customers. But the hyperscalers, I think, have done a great job creating an accelerator for customers, developers to get access to products like MongoDB and for MongoDB to be able to give them tools with great experience. So it's a great time to be together.

>> And they got to think about serving you as a partner because you guys are a big partner of AWS.

Alan Chhabra

>> We are.

>> I mean, it's been well documented. I don't need to go into it, but I want to bring up something in my article with Matt Garman and get your reaction. He said, "I want to get back to the basics." Gen AI's getting all the headlines, but he emphasized they're committed to the foundational services, compute storage and databases, is what he said. And then he said, "The theme emerging is the critical role of applications, inference, generate predictions is going to be integral for all applications." And he sees inference as foundational of a component for AWS search on par with compute storage and database. "It's the next building block, core building block," Garman stated unequivocally. What does he mean, because he's kind of saying databases aren't going away, but he's asserting that inference service, inference building block is doing the code, it's not a database, but it's like it's smart.

Alan Chhabra

>> Yeah. I like to simplify this. Think of it as developers over the years have written applications, applications for e-commerce. I mean, the Black Friday was last Friday and then Cyber Monday today. There's so many e-commerce applications written. You always need a database for real time for e-commerce applications. But in the last couple of years, what you've seen is in the world of AI, people have focused on training models and experimenting how to train models with using their own data and public data. And there's been risk. People are scared of what to do, the legal ramifications, but there's been this whole influx of training models, and then you see the uptick of people using chips like Nvidia to make that process faster. But you still have this whole plethora of applications that have been written and are being written by developers, like I mentioned, the e-commerce application, how do you combine those two worlds together? To me, that's where a company like Mongo and inference plays together. Mongo has a database that has a vector component. It can easily tie into the most popular LMs today. So if you want to bring in agentic workflows or RAG workflows into your applications either new or previously built, that's the building block I believe Matt's talking about, connecting this world of training all these models for the last year to the application suite, which is how companies make money. They make money on writing applications and selling things to consumers.

>> I think you got it right. And here's this quote that we'll see how plays out tomorrow, but his quote is, "The entire way you build generative AI applications is going to change and be reinvented." So he's saying the entire way you build generative AI applications today is going to change and be reinvented. Inference is the next core building block, period. If you think about inference as part of every application, it becomes integral just like databases.

Alan Chhabra

>> Yes, 100%.

>> I'm not saying database is going away.

Alan Chhabra

>> No. Database is the glue between the two worlds, but I would give you, there's a third vector that we don't always talk about. How do you use AI to write applications faster and more efficiently while use AI to make them smarter? So Amazon's done a great job with Q, and Microsoft has Copilot, and Google as their own code assist. So MongoDB has been training these models so that when you're writing applications, you can write them faster and easier using the co-pilots from these hyperscalers. And then once you do that, you're able to plug the two worlds together that I've mentioned using a vector database and other products, and that's the glue.

>> .

Alan Chhabra

>> Yes, a hundred percent.

>> Yeah. Again, I wanted to point to that quote because it's important as saying it's as integral as database. Now that's important. I mean, the magnitude of that statement has implications.

Alan Chhabra

>> That's why you see investors and the world saying, "Where's the true ROI of AI?" Okay, obviously, Nvidia's done great in the markets and the hyperscalers have done great in the markets, but if you ask most CIOs of large enterprise today, they'll say, "We have hundreds of experiments going on with AI," but are there mission-critical apps being run today with high ROI? Many would say it's too early. And that's why I believe you are hearing people looking for the answer that Matt just mentioned, that the inference layer is what's going to connect these worlds to give customers the confidence to take applications to production and use the models they've been training. Now, when is it going to happen? Is it going to happen in the next month, the next year? But it has to happen for the ROI to be delivered to investors and most importantly, our customers.

>> What's the news you guys have here at re:Invent? What's the big deal?

Alan Chhabra

>> Well, I mean, AWS has been our best partner for many years. It didn't start that way as you know. You and I have spoken before. I would say we've gone from competitor to coopetition to now truly a first-class-

>> the highest sponsor level, I think you guys are with the Diamond. What's the highest level?

Alan Chhabra

>> We're Emerald sponsor.

>> Emerald sponsor.

Alan Chhabra

>> We win awards now, I can't even count them. Ryan and I were talking earlier, I can't even memorize it for the interview. We are part of their blueprint competency program, gen AI competency, modernization competency, using gen AI, automotive, financial companies. These are all blueprints where AWS and Mongo help customers together. But the biggest news is back to how do you get an ROI from AI, the inference layer, and combining those worlds, customers need blueprints, best practices on how to take an experiment and make it run in production where their end customers are going to get value from it. So earlier this summer, we came out with a program called MAAP, MongoDB AI Applications Program, and that's literally what it is. It takes partners, some competitors, some more complementary, brings them together to give customers a blueprint. So you need a blueprint for how to connect MongoDB via LangChain to Bedrock running on AWS with a data warehouse layer of Redshift. We give you a blueprint to do that. And it took a long time to bring all these players together because all of us have our own agendas. The beauty of MongoDB, we're kind of like the Switzerland here, where we fit in each of these stacks. Remember our DNA, we used to be in the mean and merge stack, M was Mongo, so worked to switch a little across all these different companies, hyperscalers-

>> You guys just had great developer penetration....

Alan Chhabra

>> we bring them together. But that's our big announcement today.

>> That is huge.

Alan Chhabra

>> We're expanding on that program.

>> But also what you guys have done is you've maintained the roots of the developer while building a system that does more.

Alan Chhabra

>> 100%.

>> And I think that's where I'm looking for this AI inference layer to be like, okay, databases and all the systems around data are changing. They have to.

Alan Chhabra

>> I'll give you an example of a blueprint, and I didn't actually know this. I didn't know that training language models on audio, pictures, video, you could do it, but audio sound files was fairly difficult. I don't think it had been done before about a year ago. We've now given customers, through our program with our partners, AWS, Azure, Google, the ability to do that and then use that training of sound models to an online application using a RAG architecture. That didn't exist a year ago. So the more of these blueprints, these recipes we can give to customers, the faster they get application to production. Why is that good for Mongo? Obviously, a production application is going to consume more than a pilot. So today, we expanded on our program, which already included AWS, Azure, Google, and included Anthropic and Cohere and smaller language models like RC. Today, we added Capgemini, McKinsey, and IBM to the program, and Confluent. So we added four more partners, and they're all bringing their own flavor to these recipes so customers can get AI applications to production faster.

>> Great. And what's your goal for the show at AWS this year, re:Invent? Our 12th year doing theCUBE, we've seen the evolution. What's your plans? What do you hope to accomplish this week besides stay awake and get through everything?

Alan Chhabra

>> Well, when I get on the red eye on Wednesday, hopefully, I don't stay awake. But I would say, look, one thing about AWS is this show is a show where you don't just talk about business, you connect with friends. I've made so many friends over the years at MongoDB. I've been there 10 years, at other companies. The one show I know, except if you're Microsoft or Google, you're going to be here. So I really hope to connect with a lot of friends from the past, talk about these AI challenges, and then go solve them. If we, as companies like Mongo and other data players, can't help customers solve them, no one's going to.

>> Al, great. Thanks for coming on theCUBE.

Alan Chhabra

>> Thank you.

>> Good to see you. Good news there. I mean, Mongo making things happen as we've been covering. Again, it's a developer scenario where the value will then continue to be created. Again, just another step function, value, opportunity for developers. Of course, theCUBE, they're covering it. We'll be right back. More from Las Vegas after the short break.