Exploring the Role of AI in Video Understanding at AWS Summit NYC 2025
Soyoung Lee, co-founder and head of go-to-market for TwelveLabs, joins John Furrier, co-founder and co-chief executive officer of SiliconANGLE Media, Inc., on theCube at AWS Summit NYC 2025. The discussion centers around the innovative strides TwelveLabs takes in the field of multimodal artificial intelligence and video understanding, highlighting its impact across various industries and potential future applications.
In this insightful conversation, Lee shares insights on how TwelveLabs revolutionizes video understanding through their AI and multimodal foundation models. Hosted by Furrier of theCube, the discussion delves into the complexities of video data, the company's groundbreaking application programming interface product, and the numerous applications of intelligent video tasks within different sectors.
The discourse elaborates on key takeaways and innovative insights from the video. According to Lee, TwelveLabs breaks new ground by providing solutions to complex video data challenges, enabling more nuanced understanding and interaction with video content. It emphasizes the importance of AI in transforming industries such as media, sports, and safety, underscoring the significance of partnerships and technology integration, according to analyses provided by Furrier.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
AWS Mid-Year Leadership Summit 2025. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Register For AWS Mid-Year Leadership Summit 2025
Please fill out the information below. You will recieve an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for AWS Mid-Year Leadership Summit 2025.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
AWS Mid-Year Leadership Summit 2025. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Sign in to gain access to AWS Mid-Year Leadership Summit 2025
Please sign in with LinkedIn to continue to AWS Mid-Year Leadership Summit 2025. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
David Nalley, AWS | AWS Summit NYC 2025
In this AWS Mid-Year Leadership Summit interview, Rajiv Chopra, VP of Amazon Just Walk Out, joins theCUBE’s John Furrier to unpack the evolution and impact of computer vision in retail. Chopra shares how AWS has transformed the breakthrough technology behind Amazon Go into a scalable, edge-powered solution for partners across stadiums, hospitals, universities and airports. With over 250 deployments outside of Amazon properties, Just Walk Out is redefining how consumers shop by enabling fast, frictionless experiences without checkout lines.
Chopra details key benefits for retailers, from revenue growth to shrink reduction, and illustrates use cases across venues like Lumen Field, UC San Diego and Hudson News. He breaks down the technological architecture behind the scenes, including deep learning models, custom edge compute devices and cloud integration, and explains how Just Walk Out balances accuracy, performance and customer experience. The conversation also highlights the broader trend of digital-physical convergence and visual reasoning as a frontier for applied AI.
Watch to learn how AWS is turning real-world environments into intelligent, automated spaces – and how Just Walk Out is leading the charge in reimagining retail through innovation.
In this AWS Summit NYC interview, David Nalley, Director of Developer Experience at AWS, joins theCUBE’s John Furrier to unpack the seismic shifts underway across developer ecosystems and enterprise infrastructure. Nalley provides an insider’s view on AWS’s agentic AI advancements, including the momentum behind AgentCore, the growing utility of the MCP protocol, and the evolution of Kiro into a productivity-first developer assistant. He explores how new tools are redefining developer workflows by combining structured collaboration, context preservation and co...Read more
exploreKeep Exploring
What is happening at the AWS Summit in New York City?add
What significant changes are being observed in developer experience and the role of agents in software development?add
What insights can be gathered about the current state and trends in the developer world?add
What is the role of developers in a business, particularly in terms of problem-solving?add
What is the current focus of your team and what events are you involved in related to MCP?add
What has been the recent transformation and focus of AWS in terms of product delivery and speed?add
>> Welcome back, everybody, to theCUBE. We are here on the show floor of AWS Summit. New York City, everybody's in town. We're here for the one-day spectacular event. Tens of thousands of people are here. Of course, it's a free event, but it really feels like a mini re:Invent because there's so much that's happened. We're at the midpoint of re:Invent, but if you look at the first half of the year, it's been quite a slew of announcements. More regions, more CapEx, more product, AI infusing in all of AWS.
And of course, it's theCUBE's Media Week for Cloud and AI at our New York Stock Exchange studio. We're featuring all the leaders, startups, and executives really driving the accelerated change of AI. David Nalley is here, Director of Developer Experience at AWS. This is the AppDev meets DevOps meets the future of AI and agents. And the news here is obviously the Nova models with customization, and of course the AgentCore. And of course, the Marketplace has its own dedicated section, which speaks to the massive growth in agentic, and of course tooling, agents and tooling. Tooling is a surprise win this year. MCP, tooling offsetting models, being more efficient with tokens. Top stories here in New York. David, thanks for coming back on theCUBE. Good to see you again.
David Nalley
>> Thank you for having me back. I'm kind of encouraged that you let me come back.>> You're always a great... one of our mixture of experts as we call. You've been on in previous events, Open Source Summit, other things, a lot of action. You're seeing the cloud-native community mature, continue to mature, CNCF, KubeCon, platform engineering, DevSecOps, as we know and love, open source thriving away. But now the app layer and the AI stack portion is evolving super fast. News here is just phenomenal. Deepak and his team has been working around the clock. It looks like he hasn't slept in 10 zillion weeks. Everyone-
David Nalley
>> There's been a lot of sleepless nights for him and his team, yeah.>> A lot of news. Give us your take because now developer experience is totally changing. Just on the Marketplace announcement alone surpassed the record. I think it was the SaaS page, hundreds of big names in the agentic section and tools for Marketplace showing real activity, real products hitting the agentic wave.
David Nalley
>> Yeah, I think in many ways what you're seeing is an indication that agents are really transforming the way we think of software, the way we think of technology. And what you're also seeing is that a lot of the companies that you would traditionally think of as software companies are leveraging their expertise in the problem areas that they try and solve because generally we build software to solve problems. They're able to leverage their expertise that made them great software companies into building agents that help people solve and automate their problems away. And I'm really excited by what I'm seeing there. It's early days, but people are actually finding a lot of value in being able to drive automating a lot of that toil away with agents.>> You know, David, one of the things I want to chat with you about is just the overall developer world right now. You got vibe coding, which turns into now production code with Kiro. All right, so that's killer. But if you look at what's happened in some of the successes, there's a lot of organic activity driving, coming out of the woodwork. For example, customers are now leaning in and giving product recommendations into the coding tools because they have needs. I covered with Matt Garman on our Halftime Report that led into this event, the success of Strands, which was an accident, a good accident. It was a developer's building tool on Deepak's team to solve something else, which turns out to be a killer product. There's these silent moments that turn out to be high-impact, like Lambda was one of those, and no one really talked about it. Only the insiders knew how good it was, but then the impact. Now everyone's using that. So Strands was one, MCP, the rise of MCP protocol, very important in this way because protocols make things work. So these are organic activities, not some top-down-
David Nalley
>> So I think developers when they're at their best, and this is absolutely aspirational, the developers when they're at their best are not building things. They're solving problems. And Strands was built internally for AWS to solve problems internally. We realized, "Wow, I bet everyone else has this problem." And the response to Strands has been amazing because it solves a problem really elegantly. And I think that at their pinnacle, a developer is solving a problem and it's not "Make work," it's "huh, there's an interesting, hopefully elegant way to take care of this problem, to come up with a solution." I think that's the value a developer actually brings to a business. It's not the "I can write 500 lines of code an hour." It's "I can think about how a business problem impacts us and I can work on solving it," or, "I can think about a technical problem and figure out how we can drive value faster."
And a lot of people are pretty dismal in their outlook on what things like vibe coding mean for software developers, and I actually think it's going to make us more valuable because the actual value comes in thinking about and working on solving problems, whether they be technical or business problems, in an elegant technological fashion.>> I like how you brought up technical and business because one of the things that's coming out of some of the agent and also the AI work is that a lot of the formulation around identifying problems is coming from the business side because we are in the middle of a business transformation wave as well as an ongoing tech transformation. We already had that digital transformation, tech transformation, but it's now more acute because the AI infrastructure's now democratizing supercomputing. I mean, opening up more compute, that's opening up more horsepower for data, right? Okay. We love that story. I mean, that's theCUBE wheelhouse. But then now on the business model side where the business logic is, the data, executives saying, "We could transform our business." So now that just asks a different question, what problems do we want to solve there that are opportunities? That's where the action is because I love this problem-solving angle because the creativity of the art of coding was to solve a problem, and code is objective. There's no bias. You either have it or you don't. It's what it is. We could disagree on a design doc and a line, but then when you write the code, it's definitive.
David Nalley
>> Right.>> So you get faster action.
David Nalley
>> Well, I think that's at the crux of things. I think if you think that a developer's job is to write code, you're probably short-sighted or you're not getting the actual value of the developer. The thing that I think made folks like Stephen O'Grady describe developers as "the new kingmakers" wasn't that they could understand and read and write code. It was that they could translate a problem, they could find a solution, and then they could use technology to solve it. And almost a little bit of alchemy there where you can take this problem or this opportunity, solve for it, and drive a ton of value to the business or to a human.>> You know what's interesting is that I do so many CUBE interviews and I love this market because it's like I'm old-school, but I'm kind of a teenager at heart. I'm always curious. I love this because it's so much fun to get involved. "Tech teenagers," the saying goes. I think you guys coined that term inside Amazon, one of your folks mentioned it to me. But the conversation really is, I hear a cool, "It feels like the '90s again," and that's from people who have lived through the '90s in a structured way. So there's a lot of that going on. Enterprise is certainly hot. And it's interesting, the enterprise is hot before the consumer side, and I think the consumer's going to lag the enterprise. We're already seeing signs of consumer waves coming. So you have a systems revolution happening, and then you have young and talented creative people coming in with AI, so the alchemy of talent is binding. It's not like, "Yo, you've aged out." No, no. I went through the systems revolution. I mean, I know a lot about systems. I mean, it's in my degree, but I haven't coded in 25 years. Well, hello vibe coding. Now, what I love about Kiro is that where I was vibe coding, where I kind of abandoned, it actually now offers me the kinds of steps with agentic that will allow me to follow through on stuff that would've been harder for me. Get the web hook, get the hooks in there, get the API, so that it will go in and do and actually collaborate, build design docs, other things that were like a product manager would do, all with identity, all with all the heavy lifting under the covers. That's a game-changer in my mind.
David Nalley
>> You know, I think the real story for Kiro for me is yes, it will work me through a structure so that we have a well-defined requirements doc, we have a design doc, we have a task list, but then those are saved as markdown files, which means I can put those in my code repository, which means other people who are looking, their tools, whether it's Kiro or not, can also understand what we're trying to build, the steps that are needed, and multiple people can start working together. Now, I think I told some other folks earlier today that when I first heard about the idea of Kiro, I kind of rolled my eyes and I'm like, I told Deepak, "What are you really going to build here? Because it can't just be another same thing that's already out there because the market's flooded with it." And he talked about spec-driven development, and I'm like, "I don't know that that's a thing," until I actually put my hands on it. I was really a skeptic and didn't know if I was going to believe in it. I'm blown away. It's transformative and it's really... I have lots of, for better or worse, I have lots of meetings, which means my development time gets broken up a lot. And the structure that Kiro has added to that process means that it's not just fire-and-forget development. It's actually there's a process and Kiro will keep up with that process, even if you have to walk away for a couple of hours or a couple of days.>> And that'll help the abandonment because this is about abandonment, right?
David Nalley
>> Right.>> A difference, "Ah, I'm busy, I'll take the kids there. I'm tired. Take a nap." I mean, that's my moral because I have meetings. So again, to your point, it's right there. It really is an assistant. All right. I got to ask about the MCP because I think MCP is probably one of the biggest surprises of the year for me in a good way because it was a rallying moment around people saying, "Let's just do this. Okay? Let's get this protocol. Models are there." And so that's one surprise. I want to get your reaction to that.
Strands, I didn't even know it was coming. That was a surprise-surprise, like, "Oh, that's a good story." It's instructive. But MCP was really community-oriented, really shows solidarity around, "Let's go faster. We just agree. Salute the de facto flag, whatever."
And then the other surprise was the fact that tooling technology is emerging super fast around models to make tokens more valuable because the tokens is the premium item and reasoning's coming on. So as a developer, I have that Bedrock layer of maybe configuring, managing the models in a way that maximizes my token efficiency because cost is a problem. Why would I want to ask a model what math equations I could do with a compute engine? So you have this, that's weird, but it makes sense when you go, "Whoa, why waste a token on that?"
So you have all these situations where now engineering things are different. Now, going back to the old school, when I was in college, the degree was software engineering. The word engineering was in there. DevOps has an engineering vibe. So I think to your point about problem solving, what does that even mean? You're engineering a solution, solving a problem. You're not just developing. Developing means I'm coding, I'm building, I'm a developer builder. Okay, I buy that. But I think there's an engineering art here, and I think DeepSeq is a good sign. What's your reaction to that? Do you feel that same experience? Do you feel like we're in this, like we're building and engineering things in a way that's systematic, systems-oriented, or is that just a little bit my bias?
David Nalley
>> I think we're having a forcing function around needing to have systems in place. The scale dictates it really quickly. And I look at MCP, and MCP reminds me a lot of the Unix philosophy: one tool with one job that's pretty simple and does it well. And MCP reminds me a lot of that. It's agent-to-tool. That's what the protocol's focused on. Will it expand at some point, do things, other things? Maybe. But it's focused around solving one thing, and it's a simple, elegant fix. I think that makes a lot of people pretty happy because they see this tool that's really imminently useful, and much like a lot of the tools in Unix, I think that people are using them in strange ways. I don't know how many MCP servers you load up when you start your various gen AI tools, but I've found that I've now got this catalog of MCP servers because there's so much use that really gives me another level of functionality atop what the model was able to do. I will confess that I'm using QCLI with an MCP server for Outlook to help manage my mail. And one of the tasks that I had it do was go look at all of my sent mail and write a prompt for me, context for me for how I author messages so that if I give you something to write, you could rewrite it in my style. And that is I would've had to have gone and manually collected all of that data to then put it in to create that. I have an MCP server and I'm like, "Hey, just go look at the sent folder, look at all of my writing, and figure out how to tell me how to write like myself.">> How'd that go? Good?
David Nalley
>> Oh, it's amazing. It's amazing and it saves a lot of time. One of the things that happens a lot is I get a lot of emails that I'm expected to send out and I look at them and I'm like, "That does not sound like me. That doesn't sound authentic." I don't know that I would ever say, "Critical strategic alignment." That's not my voice.>> Yeah, that's not your linguistic mode. Right, yeah. That's a piece of... No.
David Nalley
>> It's doing a good job of taking that content, rewriting it in my style. That's not me saying anything that I wouldn't agree with, but it's just making sure it's my voice.>> You're grounding it with your signature touch-
David Nalley
>> Absolutely.... >> so you can communicate effectively.
David Nalley
>> Which in the old days, I wouldn't have sent it out anyway. I would've rewritten it myself. But the MCP, the fascinating part is not that it rewrites, but that it was able to go look at all of my sent messages without me doing any work. MCP powered that. There's a lot of MCP servers for accessing documentation and accessing a number of services to actually give you data from what's going on rather than having to rely on the model for that.>> All right, talk about your job. I know you have a really awesome job. It's super busy. You're doing a lot of development work. There's a high discovery market right now for content. People are looking at peers. There's a lot of collaboration, a lot of trust being built, new relationships are forming. I'll use the word "replatforming" in quotes, but I'm going to say that. But a lot of people are thinking, "Okay, I got to get my hands on this. I got to learn. I got to be curious, be that teenager," even though we're not in the AI-native world, but you can still stay in it by being relevant. You're seeing a lot of people jumping in and playing around. So what's your focus right now? What are you optimizing for? And if you've got events coming up, I know you've got an MCP event in New York City, put a plug in for that. So what are you optimizing for, what's some of the focus, and put a plug in for your events?
David Nalley
>> Yeah. So we've got a lot of focus right now on just experimenting. Things are changing so fast and you need time to understand what's happening, what's new and interesting. We're spending a lot of time with the MCP, and I don't think that's any shock. I think everybody is spending a lot of time on MCP. We're working on an MCP Dev Summit in London. There's another MCP community event that's happening in New York that we're getting involved with. And so there's a lot of activity here. People want to understand it. They want to build their own MCP servers, and so lots of experimentation going there. I've got folks on my team who are trying to set the bar on what should that look like? For instance, there's this, and I guess I just confessed to it, there's this, what I think of as an anti-pattern in MCP usage right now, which is a local MCP server. I really don't think that that was ever intended to exist, but we've worked around it to... We're using MCP servers because we don't have things like authorization or delegation of identity figured out. And so we say, "Oh, well, if it's running on my machine, it essentially runs as me, and anything I can do, the agent or the model can do."
And so I think in many ways, right now we're abusing MCP in that way by doing a lot of local things.>> Delegation is trust. That's basically a trust issue.
David Nalley
>> Exactly. And whether that should be or not is a different story. I don't know if you saw in Swami's keynote today where they were talking about Bedrock's AgentCore, where a lot of those things around how do we store memory of what's happened previously so that we don't have to go and grab all of that context crutch? How do we figure out identity and authorization to systems? Because MCP is really designed to be a remote protocol, not something that you're running on your local machine. And I think.>> So state, things of that nature become a challenge. Right?
David Nalley
>> Right. And one of the things that AWS is trying to do is trying to accelerate people to not have to reinvent that wheel with every single MCP server or every single agent. How do we figure out to let these agents behave in a way that is a common platform? So you're seeing AgentCore help with that, the Strands, SDK. One of the things we discovered is that people were doing a lot of reinventing the wheel every time they wanted to build an agent. And so when we built Strands originally internally, it was to solve some of that reinvention. I think that's one of the things I'm excited about is people saying, "Hey, where are the things that we're being repetitive?"
One of the best attributes that I think that developers have is being lazy, and I'm not saying that disparagingly.>> That's a good thing because they want to work effectively.
David Nalley
>> Because then they want to automate it. And so looking around the corners and seeing, "Hey, we're doing a lot of duplicate work here.">> Yeah. It's going actually to good engineers, like, "Why duplicate this?"
David Nalley
>> Yeah. They don't want to push the button five times when they can automate it. And so I think that's a key attribute in a good developer is wanting to automate it away so that you can... that laziness of not wanting to have to turn the crank continuously.>> Right. It's a motivating thing. I mean, Jensen Huang would always say on stage, like, "I want to delay decisions because maybe they'll go away, meaning people will actually solve it." So his lazy strategy is to... Or hey, be smart. Don't make it a crutch. Make it efficient. David, so much to talk about. And again, congratulations on great product coming from the teams. It's really interesting. And I will say I've done, I did a little preview, Halftime digital event. Matt, 10 executives, Atul came on, all the top people came on. And I got to say that AWS, and I'm not saying this because I'm biased, I do, I love the cloud play that you guys always had from day one, but just the transformation of AWS internally within the past year and a half at the size and scale of the company has been pretty remarkable. Because I'm reading between the lines and all the messaging, it's the deliveries and the products, I mean, the products that are shipping and the productivity is quite impressive, the speed.
David Nalley
>> Yeah. And you know, AWS and Amazon have long held that speed matters. And I think there's really this focus on this is an exciting time, things are changing every week, and if we're going to deliver value for our customers, we're going to have to do that fast. And that's definitely a focus.>> Well, I'm looking forward to doing more content with you. I know you've got a big agenda. You're putting out a lot of content. The market needs it. Just trying to look at what's relevant, what's real, what can I rely on? I want to save some time and reduce the steps it takes to do stuff and get to more beer time, as one developer told me. "I got to free up some beer time" when I asked about productivity.
David Nalley
>> Yeah, I'll toss one other plug out there. One of my colleagues, Jim Howitt, just released a blog post about how we think about developer productivity. And I think for many years that's been how many pull requests do you submit or how many lines of code? And he's got a novel measure called cost to serve that he's worked with a number of Amazon scientists.>> Really?
David Nalley
>> And so it looks at things like how much rework do you have to do?>> And it could be a .
David Nalley
>> That's a factor into quality. How many changes are you delivering? How many new features are you delivering? Because that's one measure of value, but it has a number of tension metrics because when we used to measure against lines of code, that would optimize you for certain things. So that's a great blog post. I'll send you a link to that.>> Yeah, send it over. We got to roll, but thanks so much for coming on and great to see you again. We're grateful for the work you do. And again, the community likes it as well. They need it. And thank you for coming on theCUBE.
David Nalley
>> Thank you for having me. I appreciate it.>> All right, great. We're here. All the action is in New York this week. Today it's AWS Summit at the Javits where AI is continuing to change the game. AppDev's going to AgentDev. All that platform engineering investments in DevSecOps and cloud going next level. I'm John Furrier, doing our best to bring you the action, sharing that data with you. Thanks for watching.
