"We brought the entire engineering and the science team right in front of customers because we wanted the entire team to hear the feedback from the customers first hand... bringing engineers as well as scientists into the room with customers was a game changer."
What if creating a professional video for your business was as simple as a single click?
We're joined by Kabir Bedi, Head of Product for Image and Video Generation at Amazon Ads, to discuss how generative AI is making that a reality. He provides an inside look at Amazon's mission to democratize video advertising, empowering everyone from mom-and-pop shops to large enterprises with their innovative video generator that evolved from simply showing products to showcasing them in realistic, multi-scene lifestyle settings.
Kabir shares key insights into the engineering culture that powers this innovation, built on deep customer obsession and a dynamic, experimental mindset. He explains why bringing the entire team—including engineers and scientists—into direct conversations with users was a game-changer for their development process. Plus, listen as Kabir explains why a "learn and be curious" mindset is the key to thriving in this new landscape of AI-driven product development.
Show Notes
- Follow Kabir on LinkedIn
- Learn more on the About Amazon Blog and the Amazon Ads Blog
- API Orchestration: Where Developer Agility Meets Business Impact
Transcript
(Disclaimer: may contain unintentionally confusing, inaccurate and/or amusing transcription errors)
[00:00:06] Andrew Zigler: Welcome to Dev Interrupted. I'm your host, Andrew Zigler.
[00:00:10] Ben Lloyd Pearson: And I'm your host, Ben Lloyd Pearson.
[00:00:12] Andrew Zigler: This week we're talking about meta making the news for business practices and privacy violations, Cursors own hiring practices, the joy of toy software, and how API sprawl is slowing you down. Ben, what's on your mind?
[00:00:28] Ben Lloyd Pearson: Well, I can't pass up the opportunity to highlight, all of the horrible privacy practices over at Meta. So maybe we can just start there.
[00:00:37] Andrew Zigler: Okay, this is the story that broke the news in the last week, talking about meta attempting to approach AI engineers from companies like OpenAI, uh, and doing so with huge bonuses and paychecks and, uh, sign on, agreements. Some of them topping a hundred million dollars. Could you imagine a compensation package from a competitor with that kind of price tag attached [00:01:00] to it?
[00:01:00] Ben Lloyd Pearson: My mind literally cannot fathom what that kind of money feels like or looks like.
[00:01:04] Andrew Zigler: Yeah, so obviously Sam Altman and, and, and Zuckerberg, they took it to the news and there's been a whole news cycle about this. And Altman proudly said that none of his employees took that offer. They all turned it down and his employees believe in open ai. So it obviously became, a posturing.
[00:01:21] Andrew Zigler: Conversation between the two companies. but it's really interesting to see because only a handful of companies can really play at this level. I mean, a hundred million dollars sign, a sign on bonus to attempt to poach competitor talent. What about the rest of us? there was a really good post that I saw on LinkedIn talking about this from Mohit Malik.
[00:01:40] Andrew Zigler: That said, real strength is not poaching. It's in growing, retaining and respecting the people you already have. And so that was my positive takeaway of the story of, you know, there's always opportunities to take those a hundred million dollar, poach deals and invest them internally on making your own engineers the best in the field.
[00:01:58] Ben Lloyd Pearson: the thing is that [00:02:00] Mark Zuckerberg in particular, I feel like he has a well established history of using the, throw money at it strategy to try to solve problems. So, on one hand it actually would not surprise me at all if this was true. if you look at meta, you know, they really were like an early leader in LLM Research and development, and they made a lot of great advancements in open source contributions to the cause. so I think they really do deserve some respect and recognition for their contributions, like in the, the field of ai. but it really does seem like they are struggling to build AI products that people actually want to use. You know, we've seen like Zuckerberg talking about ai, social media bots that interact with you and become your friend or. Um, you know, meta's out with their new, chat interface for ai, that some people have been using. but I, I just feel like there's, there's so many more interesting problems out there to solve with AI that like, social media just isn't, [00:03:00] it's not really it, you know? so I kind of imagine that Zuckerberg feels a little bit behind in all the AI hype. When I, when I really dig into this, it feels like something's off. Like, if this is true, it's, it is truly a ridiculous sum of money. Like
[00:03:13] Andrew Zigler: Truly.
[00:03:13] Ben Lloyd Pearson: it feels, if it's true, it feels like an attempt to disrupt a competitor rather than any sort of real investment into ai. I personally would not trust this offer unless there was like some sort substantial compensation that,
[00:03:27] Andrew Zigler: Yeah.
[00:03:27] Ben Lloyd Pearson: that like meta revokes this offer suddenly. 'cause I feel like that's a real potential here. but you know, honestly, I just don't really take the personal drama of random tech billionaires too seriously, because, at the time of this recording, Zuckerberg hasn't responded to any of this.
[00:03:41] Ben Lloyd Pearson: So it's just the conjecture of a single person. And Altman could be playing his own games here. So, but I will say this, if I were an open AI employee and I heard that my peers were being offered this kind of money, I would start wondering, Hey Altman, am I also worth that much?
[00:03:58] Andrew Zigler: you know, you might turn up your [00:04:00] nose at the Real Housewives of Silicon Valley, but I'm definitely gonna stay tuned to the gossip and we'll let our, uh, listeners know if any new developments happen.
[00:04:07] Ben Lloyd Pearson: Yeah, so I mentioned the meta privacy stuff, so let's dive into that too. 'cause let's just keep this, meta hate train going a little bit.
[00:04:14] Andrew Zigler: Well, you know, it's a nuanced hate train, but this next part of it is quite interesting. 'cause I think that if meta convey a hundred billion dollars for signing bonuses, they can probably handle whatever fine they're gonna get for breaking GDPR and DSA and DMA and all the other alphabet soups of privacy enforcement out in the world.
[00:04:31] Andrew Zigler: Because meta,
[00:04:31] Ben Lloyd Pearson: there, there was a, a news broke, another one this week about them, about how they Dev devised an ingenious system called local host tracking that bypassed, even things like Android sandbox protections that protect your identity while browsing in incognito mode. And incognito mode is something that we all take for granted.
[00:04:48] Andrew Zigler: and obviously there's always a, a way for your identity to be compromised through while using incognito mode. but how this one worked is, Basically [00:05:00] your Facebook and your Instagram apps, they would set up like a hidden, kind of like a intercom basically inside of your device.
[00:05:06] Andrew Zigler: It was a port that would listen, and so when you would use incognito mode and you would go to websites that had a tracking pixel, that tracking pixel would actually reach out on your local device to that open port and then use that to identify you on Meta's server. So this effectively bypassed all of what.
[00:05:22] Andrew Zigler: Incognito mode is intended to do, because incognito mode protects you from yourself. It,doesn't save anything on your own device. It works on your own device, right? But this works by actually, sourcing meta's, tracking pixel from the content. you navigate to by any other definition, you would call this like a virus or a, a bypass, right?
[00:05:43] Andrew Zigler: It's doing something that you're not, that is not, uh, a visible to you on the surface. So that was really interesting development that came out this week about how some of the underpinnings of their own software work, I'm sure there's probably gonna be more conversations about this massive privacy violation.
[00:05:58] Ben Lloyd Pearson: Yeah, you, you know, this [00:06:00] used to be a controversial take, but I think it's becoming less and less. you know, I deleted all meta apps from my phone more than a decade ago, like long before it was even called Meta. and I never looked back, to be honest. they have a very long history of just abusing and violating privacy requirements. and even if they're not breaking the law there, there's still definitely many examples like this where they're doing things that are, at least somewhat hostile to their user base.
[00:06:25] Andrew Zigler: Yeah. Yeah.
[00:06:26] Ben Lloyd Pearson: my advice is honestly, just don't trust them with access to something as sensitive as you're mobile devices. And, I, I do have a bit of a hot take on this. I, I actually have this theory that if you just install the apps from four different companies off of all of your devices, your life will get significantly better. So meta is at the top of that list. I put byte dance on this list as well. I think X deserves to be there and I've got a fourth.
[00:06:50] Ben Lloyd Pearson: You wanna take a guess, Andrew, what that might be?
[00:06:53] Andrew Zigler: So, just so I understand, these are the four apps that you should uninstall from your phone in order to have a better life.
[00:06:58] Ben Lloyd Pearson: should install, all [00:07:00] uninstall all of their apps.
[00:07:02] Andrew Zigler: Well I have to brainstorm for a second 'cause I'm like, what's the company behind Merge Mansion? Because definitely uninstalling that one is gonna give you a lot of hours back.
[00:07:09] Andrew Zigler: At least if you're me. but Hmm. knowing you, Ben, definitely Reddit, that's another one. I think you would, you would delete.
[00:07:15] Ben Lloyd Pearson: gone back and forth on that one a lot actually, it is currently still on my phone though, so.
[00:07:20] Andrew Zigler: Oh, okay. So we don't know. Maybe that's even the secret fifth one we haven't gotten to. What is it? What is it, Ben?
[00:07:25] Ben Lloyd Pearson: PDD holdings.
[00:07:27] Andrew Zigler: What is that?
[00:07:29] Ben Lloyd Pearson: so they're the, the makers of Temu, you know, it's, I
[00:07:32] Andrew Zigler: Oh,
[00:07:32] Ben Lloyd Pearson: feel like that
[00:07:33] Andrew Zigler: okay.
[00:07:34] Ben Lloyd Pearson: instant gratification style app, like get something
[00:07:37] Andrew Zigler: Oh.
[00:07:38] Ben Lloyd Pearson: whether it's like social validation or a piece of clothing from China, like, you know, there's a lot of apps out there that, that are like this, that, uh, meta is not the only company, I guess is what I'm trying to make.
[00:07:48] Ben Lloyd Pearson: They're just one of the more egregious violators
[00:07:51] Andrew Zigler: Yeah, and, and after you uninstall all of those apps, then you have plenty of room to install Roblox, which I recommend as being on everyone's phone.
[00:07:58] Ben Lloyd Pearson: Yeah, absolutely. [00:08:00] But you know, more importantly, I think this is a yet another example of how US regulations really have not kept up with technological advancements. Like we have this expectation of privacy in the physical space. it's built into a lot of laws here in the us uh, and even around the world in some places.
[00:08:17] Ben Lloyd Pearson: But we haven't really adapted this concept to the digital space as well. So as a result, we rely on these US based companies primarily to become like the defacto policy enforcement organization and not
[00:08:29] Andrew Zigler: Yeah.
[00:08:29] Ben Lloyd Pearson: US. Like it often, like their reach extends into other parts of the world as well, and without any sort of like adequate privacy protection in place like companies like Meta get to decide how private data should be handled. And you know, you mentioned that there are some potential European data regulations that Meta has violated as well as some others. I'm sure this is not the last we're gonna see of this story. but you know, I just wanna say the privacy researchers who discovered this like deserve kudos for this work.
[00:08:59] Ben Lloyd Pearson: Like, I really [00:09:00] love when, when people uncover this type of stuff and share it with the world. So, great job to them.
[00:09:05] Andrew Zigler: Yeah.
[00:09:06] Ben Lloyd Pearson: So, all right, Andrew let's talk about Cursor now.
[00:09:08] Andrew Zigler: Uh, yeah. So cursor's a hiring playbook discussed a bit in the news. I think it's just worth calling out for just a, a brief moment because as everyone knows, cursor's been under like a magnifying lens from the rest of the industry. Everyone's been trying to figure out how this, like ARR hyper unicorn grew from a hundred million to 300 million a RR in just six months, right?
[00:09:29] Andrew Zigler: So everyone's looking closely at fundamentals of a company like Cursor. So. In the last week, we had a closer glimpse at how they hire their engineers, and it's a good look into how an AI first company is looking at evaluating new talent. So on day one, these candidates are onboarded, within the company.
[00:09:46] Andrew Zigler: The engineering orientation around how they work, they're briefed on a project they'll be working on together with a staff member and introduced to the team's workflows, tools, and goals. So effectively, you, you come in to shadow, to work, uh, [00:10:00] when you kick off your interview process, and they actively contribute.
[00:10:03] Andrew Zigler: Towards brainstorming solutions, attending meetings, and they effectively collaborate as a full-time member on this by this, end of the second day. And that entire process, it has a feedback loop where they're constantly talking with their hiring manager with their. potential future teammates about the process and what they're working on together.
[00:10:23] Andrew Zigler: So it's actually really fascinating to think that just for interviewing and for, bringing in new talent, that they have such an open door on how they work and what they're working on and when they're evaluating new talent. So that was something that really stood out to me. The other thing that really stood out is that at no point in this conversation is there an arbitrary.
[00:10:42] Andrew Zigler: Take home exam or a coding leet code test or any other evaluations because these engineers, they do code while they're there, but they're more focused on evaluating their communication and problem solving skills.
[00:10:56] Ben Lloyd Pearson: Yeah, and I find it interesting that they don't allow AI in any [00:11:00] of their interviews, which is the type of thing that to me, really feels like it only works for companies like Anysphere, which is the company behind Cursor, you know, they don't necessarily need to test candidates on AI skills because it should naturally come out within the interview process.
[00:11:15] Andrew Zigler: Yeah.
[00:11:15] Ben Lloyd Pearson: what I mean by that is, you know, Andrew, you and I, we use AI a lot here at Dev Interrupted. Now, like it's really unlocked a lot of new workflows for us. It's accelerated a lot of routine work. it's been just a huge change to how we get things done. And as a part of that, you know, we've adapted new language, new practices, new tooling. we've become familiar with this whole, like, new space of operating, so if we were going to interview somebody to join this team, for example, I think it would become pretty relevant to us or pretty obvious to us pretty quickly, whether or not that candidate has, you know, sort of built AI into their lives because they would either have the same understanding the language that we use, or they wouldn't. And if they didn't, then you know, it would be, it would come [00:12:00] out pretty easily without having to like ask them to use AI as a part of the interview. Interview for example.
[00:12:06] Andrew Zigler: Yeah.
[00:12:07] Ben Lloyd Pearson: but, uh, what I really, really love about this is there's, I. So many stories out there about AI taking over the hiring process. It's actually like really comforting to see an interview story that is distinctly human. Like they're focused on the human side of the, of the, the, the experience. Rather than like, can you produce more code with AI kind
[00:12:27] Andrew Zigler: Yeah,
[00:12:28] Ben Lloyd Pearson: So it's pretty cool to see.
[00:12:30] Andrew Zigler: totally.
[00:12:31] Ben Lloyd Pearson: All right, what's next for us, Andrew?
[00:12:33] Andrew Zigler: Yeah. So, um, uh, you know, I mentioned at the top of our new segment that running toy software is a joy, and there was an amazing article I read last week from Joshua Beretto. as someone myself who was a hobbyist. Dev turned, engineer, like many of the these joyful game Dev projects he listed out were things that I certainly tried to do or worked on as part of building my problem solving skills with code and experimenting with one-off projects.
[00:12:58] Andrew Zigler: And so he gives a recap in this [00:13:00] article, some of his favorite coding projects, and he gives a list of ones that he recommends any kind of hobbyist. game developer or just code writer would try. And what's great about these is there's so many parallels in them to solving high level problems and software engineering that when you look at these steps, you really do see how the evolution of someone's skills can evolve through projects.
[00:13:19] Andrew Zigler: And there's a few in here that were really fun to me, that I've actually done before, like, Someones I'll call out or like a, a, a RegX engine, like I've written one of those before. A physics engine who hasn't, if they've ever done game development, uh, your own text editor definitely hacked around with that.
[00:13:34] Andrew Zigler: A rasterizer, you know, I love working with shaders in the, in the browser. uh, and the last one that really made me laugh was, uh, an ECS framework. Which for those that don't know, is entity component system. And uh, really that's like something that a lot of hobbyists a game devs aspire to have is a perfect version of one of those systems.
[00:13:54] Andrew Zigler: So, great list. highly recommend you all check it out.
[00:13:58] Ben Lloyd Pearson: Yeah, it's pretty awesome. Uh, [00:14:00] yeah, I really love the personal project. I am gonna disagree with the author on one key point that they make that, like, I think it somewhere in this, they make a point to not use AI for these projects. And I actually have the complete opposite perspective on that.
[00:14:13] Ben Lloyd Pearson: Like almost relearning how to code. 'cause I've sort of been outta the practice for a little while, just with the, the type of work that I've been focusing on. But with all these new agentic AI tools, I felt the need to get back into it and to start, not really learning from scratch, but in many ways like, starting from you know, not square one, but like square two or three in many cases. and you know, I think the important thing is like, yeah, when you let cursor or copilot do everything for you, you're not gonna learn anything from that. But it can also be an immense help when you are actively trying to learn. So, you know, I think my next experiment is actually going to be working with like cursor rules that help make cursor more of a guide that helps me solve problems rather than it writing all the code for me.
[00:14:56] Ben Lloyd Pearson: Uh, because, I have really been focusing on like, the fundamentals [00:15:00] with this because I feel like that's important when, you know, we're adopting this new technology
[00:15:05] Andrew Zigler: Totally.
[00:15:06] Ben Lloyd Pearson: Yeah, so what's our last story, Andrew?
[00:15:08] Andrew Zigler: There was a, a new report that came out recently that actually it dropped today. and I've been looking at it and really what's fascinating about this report from Apollo GraphQL is it looks at the state of API orchestration with an organizations.
[00:15:21] Andrew Zigler: And there were some really profound stats that I saw, this morning when it came across my desk, which is why I included it in our discussion. Things like 76% of enterprises can't make quick changes due to rigid infrastructure. Like, come on. That's something we talk about on here all the time.
[00:15:36] Ben Lloyd Pearson: I
[00:15:36] Andrew Zigler: Less than,
[00:15:37] Ben Lloyd Pearson: at all.
[00:15:37] Andrew Zigler: yeah, less than 30% can test their software without cross team coordination.
[00:15:42] Andrew Zigler: That's a reality within many teams. You know, you get these disjointed workflows, and then 52% of.
[00:15:48] Ben Lloyd Pearson: it is also a commonly sorted bottle or cited bottleneck for companies that are adopting ai, right? Like they still have these testing workflows, they get bogged down.
[00:15:57] Andrew Zigler: Oh yeah, absolutely. Like if you generate a lot of code, then you're [00:16:00] just create more, uh, throughput later on in your pipeline. You have to have the, we means to support that. we've talked about that extensively even in our webinar. We had recently another one here too, was that 52% of organizations struggle with large scale code changes.
[00:16:13] Andrew Zigler: That's a reality. You know, large companies are having trouble, doing this digital, or rather this like, AI transformation and, and, and modernization of their, of their software, their infrastructure to handle how technology is being built now. So, I, I really recommend everyone checking out the state of API orchestration report.
[00:16:31] Andrew Zigler: I'm gonna include it in our, in our newsletter. Uh, and it's something that helpful to keep in mind that as your work. Load scale, you know, this kind of thinking of orchestration, it's not just helpful, it's mandatory. And there's a lot of research in here that highlights, how you get these per performance gaps if you don't prioritize it.
[00:16:47] Andrew Zigler: So I know our audience loves these kind of data deep dives. We make them all the time. We read them and cover them here all the time. So if you wanna read this report, we'll include it in the newsletter. And as a final tidbit here, I'll leave you with is that. [00:17:00] We're actually bringing the CTO of Apollo onto Dev Interrupted an upcoming week, so stay around for Matt Debergalis
[00:17:07] Andrew Zigler: He's gonna be coming up, later this summer. Uh, so stay tuned for that one.
[00:17:11] Ben Lloyd Pearson: Yeah, it'll be really great, to learn from Matt. and what I really love about this is it, it actually aligns with a lot of the research we've been doing here at Dev Interrupted. in partnership with LinearB, you know, as I mentioned, there's lots of bottlenecks outside of code production that AI really is not unblocking yet, and I think it's really important that engineering teams just don't rush off to generate more code without fully understanding.
[00:17:34] Ben Lloyd Pearson: Impact of it. you know, we just published this six trends in AI driven software development guide. we cover a lot of the biggest bottlenecks that teams are facing when driving AI adoption across the organization. And they align quite well, with what we're seeing in this report. So, our guide is also full of some really great advice on how to address it as well.
[00:17:53] Ben Lloyd Pearson: So check the show notes. We'll, we'll share a link down there.
[00:17:57] Andrew Zigler: Yep.
[00:17:57] Ben Lloyd Pearson: So, Andrew, tell me about who our guest is this week.[00:18:00]
[00:18:00] Andrew Zigler: Yeah, this week I'm sitting down with Kabir Bedi and Amazon ads, and we're taking a look at what video generation can mean for mom and pop businesses, how AI engineering culture works at AWS and how Kabir's team built an AI product and brought it to market while the fundamentals constantly kept evolving around them.
[00:18:19]
[00:18:21] Andrew Zigler: AI is creeping into every part of the SDLC, but how far have teams really gone? LinearB surveyed over 400 devs, many of them like yourself and the Dev Interrupted audience and found that 67% are already using AI to write code, but is that creating opportunities or bottlenecks?
[00:18:39] Andrew Zigler: Our new DevEx guide breaks it all down, including adoption patterns, pitfalls in the AI collaboration matrix that charts your own team's journey with ai. Check out the guide and maybe even the panel I hosted about it with Atlassian, AWS, and ThoughtWorks, I'll drop the link in the show notes. I.we're [00:19:00] diving into how technology is reshaping video generation and how generative AI is a huge part of that evolution. We've seen some amazing developments recently, and today we're talking with Amazon. We're joined by Kabir Bedi, the head of product for image and video generation at Amazon Ads.
[00:19:18] Andrew Zigler: And we're gonna understand how generative AI is being used at Amazon and what it means for the future of generation, and also how engineering teams have come together to build this new technology that's opening up a whole new door to video creation for people just like you and me. So Kabir, thanks for being here to dive into some of the advancements coming out of your team.
[00:19:41] Kabir Bedi: Thank you so much, Andrew. It's great to be here. I'm very excited for our conversation and to discuss the latest, AI launch that we just had, which is an enhanced video generator for brands to create videos.
[00:19:54] Andrew Zigler: You know, video generation is a really hot topic right now. Everyone is, is kind of tackling [00:20:00] this topic, both with their new own technology and trying to use it in a consumer space.
[00:20:04] Andrew Zigler: So. There's a lot to explore and to learn from. And for the folks listening to Dev Interrupted, we're gonna see a lot of parallels in this conversation to, things in our own world in software engineering and how tools and things available to us are changing and democratizing the way that we work. So Kabir, if you were to give us like, a big overview of some of that announcement, what were some of the biggest things that stood out to you, and, and what role did your, did your team play in bringing those to fruition?
[00:20:32] Kabir Bedi: Absolutely. So let me take a step back and first talk to you a little bit about what's the general problem that we were, we were working to solve You said it very, very rightly. Right now, video is everywhere. Video is super important. We interact with video on pretty much most platforms that we all are, are on. Same goes for video ads, but the challenge that, advertisers face is that video ads are really expensive [00:21:00] to make. They take time, energy, effort, money. And so as a result you have so many brands that were just not able to create video ads. So we, we set out on a mission to democratize video advertising so that all types of brands, whether You're a mom and pop shop that's selling one product on Amazon or a large brand selling hundreds of products are able to use video ads. We launched video generator in beta last November, and over the last six months have been receiving really good feedback from advertisers and just now launched an enhanced version of it with a bunch of new capabilities that I'm super excited to talk about that I think will really. Meet the need of those advertisers that have never been able to try video advertising before because it was too costly or too time consuming. and, you know, using the power of AI to [00:22:00] really sort of bring their products to life in a way that they've never been able to do before.
[00:22:04] Andrew Zigler: Yeah, this is like a great introduction to the problem and I like that. Zoom in on it, being accessible to, you know, more mom and pop shops and places that didn't traditionally dive into huge scale asset creation and having the photographer come out and. Set up all of the stock imagery backgrounds or to, you know, to buy all of those stock images online.
[00:22:26] Andrew Zigler: It like that, there's a huge amount of effort that goes into constructing creative works for that type of business now to make it more accessible to them, but without sacrificing quality. But it allows them to kind of open the door on how they position their own products and right out the gate too, you pointed out something that like stood out to me as a, a really big parallel is just making it accessible to.
[00:22:49] Andrew Zigler: As many folks as possible. you see the same thing happening right now in like engineers that are adopting AI tools. it makes and coding like vibe coding, as you might call it, more [00:23:00] accessible to more folks to try out, to build things for the very first time.
[00:23:04] Kabir Bedi: Absolutely. And I think that, you know, we had heard that consistent feedback when we launched in beta that we love that this allows us to create a video that we've never been able to make before, and our most recent data really validates that story. You know, what I've seen is that over half of the videos that are now getting created. video generator are for products that have never been advertised using video on Amazon before. so right off the bat we're seeing that advertisers or brands that have not been able to advertise their products using video are using this tool right away And with this launched, we launched a bunch of new enhancements that really have sort of gotten folks excited. One of the pieces of feedback that we heard when we launched in beta was that back then that technology was, was really cutting [00:24:00] edge, was to show how products looked in a video. but the feedback that we were getting from brands was that actually, I want to go beyond that. I want the video to show how my products will be used instead. And so that's one of the big, big pillar innovations that we've just launched, which is we've added a lot more realism that now shows products in use. a great example could be, previously we could show how a watch looked on, say,
[00:24:28] Kabir Bedi: a table. Now we can show how that same watch looks on a human's hand, showing how the human moves their wrist to see the time. So previously showing things on a flat surface really worked well for certain categories like kitchen or say home. But for things like apparel or, electronics wanted to show how their products are gonna be used.
[00:24:52] Kabir Bedi: And so that's, that's sort of one of the pillars of innovation that I'm super excited about. The other one is also, you know, as [00:25:00] we've. invested in innovated in new technology. We've been able to create multis scene ads. So now you don't have to have just an ad that has a single scene.
[00:25:09] Kabir Bedi: Now you could have three to four scenes that makes the ad, a lot more comprehensive to show you really the product from all its different angles, all its different use cases. And so you're seeing this is sort of latest, you know, I would say greatest in technology. Similarly, plenty of other innovations like the ability for advertisers to now upload an image and convert that into a video or upload a video and have us summarize that, by actually identifying the key frames in it to create a video ad that suits that brand. So a bunch of new innovations have come out and we're getting very strong feedback, you know, I'm very happy about.
[00:25:45] Andrew Zigler: that's an interesting way of tackling the problem. And it actually brings me to one of my big questions I wanted to tackle in this conversation with you is about creating this tool, you know, and, Really changing the level set right in which folks can actually approach, uh, video [00:26:00] asset creation.
[00:26:00] Andrew Zigler: But in doing so, you're getting a whole new generation of folks who've never used this tool before. So you have to think of all of the different ways they're gonna interact with it. And this one thing that you hit off the bat is really fascinating. This like, multimodality of folks being able to input like a video or input an image or input like whatever.
[00:26:19] Andrew Zigler: They had. Right. So what kind of assumptions did you have to throw away about how maybe your traditional audience used these tools? More tech savvy audiences, in order to make it something that would be massly adopted by other folks.
[00:26:32] Kabir Bedi: Yes, yes.
[00:26:34] Kabir Bedi: So, one of the, the sort of assumptions that we worked backwards from was that we have to make it easy to use, which means how can we reduce the amount of input? That's really needed from the user to give them a high quality output. So that's why one of the features of this, of this product is that it's one click, which means that all the advertiser provides is what product they want to advertise.
[00:26:59] Kabir Bedi: And [00:27:00] we handle everything behind the scenes, which means prompting, generating the video, et cetera. And, I was recently on an advertiser call where they really admired that, this was an advertiser that had been playing around with, with, you know, generally ai and, concluded that they felt that AI required a lot of high quality prompting to get a high quality output. And he told me that he loved, what he loved about this tool was that he didn't have to do any of that. He said that you've made it so simple for me that I don't have to think about what should be my input prompt, what should be the guardrails and the requirements for me to get a high quality output. And so that is, I would say, one of our biggest, pillars that we really sort of wanted to hold true, which is how do you make it easy? Because a lot of the brands, smaller mom and pop brands haven't, like we already discussed, haven't created video before. They also
[00:27:54] Andrew Zigler: Right.
[00:27:54] Kabir Bedi: with AI before. And so you want to really make it simple for them by [00:28:00] actually reducing the amount of decisions that they have to make to also reduce their cognitive effort involved.
[00:28:06] Andrew Zigler: you're really driving now at like, the core of what I think makes this problem solve so fascinating is it requires this whole new mindset of innovation to build a tool that's gonna be used at the scale by this broad of an audience. I think there's a lot of folks that, that folks can learn from how maybe you approach solving that problem.
[00:28:25] Andrew Zigler: So what were some initial things about, bringing this to market that. Maybe like go against the grain or you had the unlearned traditional mindsets to, to get it out the door. you're building on a technology that's so as innovating so rapidly, like we've heard now all the things you've walked through that it can do and all the modalities in which you can interact with it.
[00:28:45] Andrew Zigler: And it's like nothing short of impressive. Right. And we're talking about. Video generation. So what has it been like to innovate in that category as it's almost been like, uh, speeding ahead with you as well, in terms of just like every day there's something new.
[00:28:59] Kabir Bedi: [00:29:00] Absolutely. I think that the, that the rate of progress is so inspiring that everyone working on this, including myself is you know having, having so much fun in the process.
[00:29:11] Andrew Zigler: Yeah.
[00:29:11] Kabir Bedi: along the way I've also felt that yes, the way that we have to build these AI products, particularly this sort of product has evolved, over time. two key things come to mind. The first one is, you know, that, when we launched in beta, what we did is that we really brought the entire engineering and the science team right in front of customers because we wanted the entire team to hear the feedback from the customers firsthand. When we are creating a video for a brand, you know, it's a high responsibility. You are showcasing their product, you are bringing their brand to life. And so it's very important for everyone on the team to, be hearing about the nuances in. In, you know, the advertisers as feedback, oh, I like the [00:30:00] way that the video did X or, you know, I didn't like that, that it sort of prioritized this aspect of the product rather than that aspect of the product. and, actually one great example comes to mind to me. We met a baby brand. they made baby books and, essentially two fathers, new fathers, that didn't like the books on offer, when they became fathers. So they chose to create this brand out of Ohio. And they had never created video ads before. largely because one, it was too expensive to create video ads. And two, they felt that babies didn't make for, for, you know, very good models. And so in this interview with them, he really spoke about how using video generator.
[00:30:39] Kabir Bedi: they had never thought that they would be able to create video ads till they use this tool. and you know, that feedback really sort of lit up the team. Everyone got excited about, what we're doing. We also got some, some very valuable feedback of, how we can improve things. you know, how they wanted to highlight certain aspects of the product. [00:31:00] and so bringing engineers as well as scientists into the room with, customers was a game changer. So that's one. and I think that permeates to all types of, AI development wherein everyone working on it must have their ear very closely to the ground. Luckily, this also matches a, you know, sort of very well into one of Amazon's, most important leadership principles, which is of customer obsession. The other area that I think the team culture has really, evolved is around experimentation, iterative development around model improvements.
[00:31:35] Kabir Bedi: You know, all of all of these AI models are just progressing so rapidly that the team has sort of now developed a culture where we do multiple experiments a week. we're just constantly trying either different models or we're prompting differently to see the output. And so it's, it's an environment which is a lot more agile, a lot more iterative and you know, I would say [00:32:00] dynamic and those are the two aspects that I have felt for any team wanting to develop. has been important shifts for us, which is bringing everyone on the team closer to the customer, as well as viewing the work a lot more iteratively, because such is the rate of progress in the model development itself.
[00:32:20] Andrew Zigler: That's really well said, and it, it also echoes what a lot of leaders come onto this podcast and say about their own engineering teams. Those are two things that are, that really help you stand apart and help you get ahead in this kind of environment that impacts driven engineering, where you're customer obsess.
[00:32:36] Andrew Zigler: And you're, you're the one smashing and building things together, but you're also talking with the folks using it and understanding their pain points and where they want to go with it. That like, it, it really helps developers build the best possible software, which is what we're all here to do, right?
[00:32:50] Andrew Zigler: We want to build really cool stuff. We want 'em to be like really impactful. and we want it to work for the folks that are going to use it. So like that key thing of. Bringing people really close to [00:33:00] it. The, even like, I like how you included the scientists in that, the folks who are manipulating, looking at the data, trying to understand the patterns and like think of it all.
[00:33:07] Andrew Zigler: They also need to be closer to those users and to how their technology is being used. Right. So.
[00:33:14] Kabir Bedi: Absolutely. Let me give you, another example of that. You know, like I mentioned already our enhanced video generator, one of the key features is that it can now show products in use. And that was feedback that we got from, advertiser interviews wherein. Advertisers would share with us that, you know, I want to see my product, how it's used rather than just how it looks. it's the scientists that heard that feedback firsthand and then evolved our technology. Previously, we were using out painting technology, which
[00:33:49] Andrew Zigler: Mm.
[00:33:49] Kabir Bedi: cutting edge, when we first launched with it. And they then evolved and now we use subject driven generation technology, which is able to show products in [00:34:00] use. And I think that the only, way that we were able to make this leap was really to help the real folks, scientists who are actually building it, trading on these models, hear that feedback firsthand that, the advertisers, for example, wanted to show a shopper drinking coffee from the mug rather
[00:34:19] Andrew Zigler: Right.
[00:34:20] Kabir Bedi: seeing the mug on the table.
[00:34:21] Kabir Bedi: And so I think that, you know, bringing folks into the room has really allowed us to really unlock various levels of innovation, including just at the most fundamental, model improvements as well.
[00:34:32] Andrew Zigler: That's such a great story to back that up. And I, I love too that it, it kind of connects all the dots there. What I love too is, is that it really brings in that. Second point that you made That it stands apart really, in terms of having a culture that, that creates innovation, that sustains this constant experimentation that rewards folks for trying new things and sharing in the open.
[00:34:50] Andrew Zigler: That's really critical. Like my own team does that. We're trying out new workflows every week for this stuff. I know a lot of our listeners are as well. And having that culture of experimenting in the [00:35:00] open is then what allows folks like those scientists who get close to the customer conversations to actually then try to figure out how to get to the destination ultimately.
[00:35:10] Andrew Zigler: Because I'm sure that shifting that entire model flow from out painting to something else took a huge, effort. And it all started with experimenting. So, really goes to show how, rewarding and supporting that experimentation and culture is important to innovating in the space right now.
[00:35:26] Andrew Zigler: and what parts of your development process have you had to bolster up to really support your team? Like you're talking about introducing a lot of new workflows and tools and constantly experimenting and change to change things up. What were things that you and your team did to kind of keep developer morale high?
[00:35:42] Andrew Zigler: To keep the developer experience good through all of that transformation.
[00:35:46] Kabir Bedi: Yes, one of the things that I learned through this process personally is that the team is also eager to hear, from the end user. That is a great source of motivation and morale boost when the team is hearing [00:36:00] from the small brands about who could never produce videos before suddenly being. able to do. So that's a huge morale boost. One of the other, avenues that we've really gone deep on that in my view has worked really well for the team is that the entire team as well as leadership on the team, is very deeply involved in the experimentation. Which means that if, if we are testing a new model and there's an output from that model, I will also dive deep to really see you know, so what is that output looking like? Where is it doing well? Where is it not doing well? it's in some way been a source of, You know, an equalizer that everyone on the team is quite involved into the details regardless of, you know, what they're working on, what level they're working on, which means that the whole cycle of sort of iteration becomes a lot more robust because everyone's giving feedback. Everyone is involved into the details and that's been super fun for me as well. I love building [00:37:00] products and I also love sort of diving into the details of, of, you know, how are things working under the hood where they're doing well, where they're not doing well.
[00:37:07] Kabir Bedi: And so I've also personally really enjoyed working very closely one-on-one with scientists and so those are the two things that I would say, bringing folks towards their customers is very motivating. And everyone being involved in the details is also very, inspiring to the folks, like the engineers and the scientists.
[00:37:27] Kabir Bedi: So they know that you care very deeply about, what they're doing.
[00:37:31] Andrew Zigler: If a software engineering leader is listed. To this and they're thinking, oh, I really need to create this culture where my own engineers can get closer to my customers and can listen to how they use my software, my tools. What advice would you give to that leader to start bringing those conversations closer together in a way that doesn't disrupt.
[00:37:51] Kabir Bedi: The techniques that, that have worked really well, for us for this product launch is that we have multiple times a week folks talking [00:38:00] to various brands, and those meetings are open to everyone on the team, whoever can join, know, sort of based on the calendar is sort of welcome to do so.
[00:38:10] Kabir Bedi: and so a very simple way of doing it is to have a sort of recurring, mechanism that constantly gives folks the option, and then not have an expectation that everyone has to join every time, but
[00:38:23] Andrew Zigler: Mm.
[00:38:24] Kabir Bedi: they have the option to do so. And, actually what I've noticed is that everyone finds so much value from those calls themselves that, they choose to attend. That would be my feedback, and advice Is to have a mechanism that gives regular touch points, for the entire team. Open that up to everyone. if you want certain folks to guide the conversation with your customer, that's okay, but allow everyone the opportunity to actually listen in. The other thing that often my team does, which works really well, is just going in a little prepared for what you wanna talk to the, Customer about sometimes, we wanna talk about the multi scene that we [00:39:00] just generated. Other times we may wanna talk about, the one click feature and you know what they think about that.
[00:39:06] Kabir Bedi: And so if you go in a little prepared, it allows folks to be in the right head space of how to navigate the conversation and you know what they can learn out of it.
[00:39:15] Andrew Zigler: Yeah, it's about treating it like another workflow, another work stream. Like you. First off, you need to have the, open dialogue conversation. That means that you need to start one, then that's the case. If it means that you need to include your engineers in those conversations as an optional, then that seems the way to go.
[00:39:30] Andrew Zigler: But creating a regular loop that keeps those two entities close to each other, lets them communicate. and then obviously creating this culture that rewards dipping your toes into that work stream to understanding and getting obsessed with that problem and what you're building to solve it.
[00:39:46] Andrew Zigler: 'cause it's a real morale booster. I know that if you make it available, folks will want to take advantage of it. So, you know, that's really, that's really good advice. And I'm wondering too, when, when y'all started on your journey to build video generator, and it's more in the [00:40:00] early days of building out your, your early beta and like getting a, uh, some traction on the idea, what were some of like the early signals that you got from these kind of customer conversations that like stood out to you?
[00:40:12] Andrew Zigler: Are there any that like really painted the road that, that brought you to where you are today?
[00:40:16] Kabir Bedi: Yes. I think two pieces of feedback that we very consistently heard, you know, sort of allowed us to continue going down this path with high conviction. One was that, It was fairly apparent that brands wanted a feature that was easy to use. I already gave you the example of, you know, what we heard from the advertiser to not have to do any prompting, but the idea to make it easy to use so that Neither the tech nor understanding AI were a barrier for all of these mom and pop brands. And so we really held onto that principle because we heard it over and over again in our feedback. think the other piece of feedback that we heard, which is also what I mentioned around that sort of helped us [00:41:00] move from outpaint to subject driven generation, was that effective videos are those that show how a product is used in its natural setting. And I think that when we heard that feedback consistently over time. It sort of reaffirmed our belief that that was the direction that we wanted to go in. So with this launch we've been able to, hold onto both of those principles. Added a bunch of new features, like I mentioned, multi scene generation, live images, et cetera.
[00:41:31] Kabir Bedi: And, uh, you know, are very happy with the results because the other sort of data point that I've been looking at is that what is the reuse, uh, of the tool? And we've seen that over a third of advertisers that are using this tool are now using it to create two or more video campaigns. And so we know for a fact that really sort of making it easy for them and showing their products in use through, you know. Through a multis scene video. It's sort of [00:42:00] actually encouraging them to use the tool more to create more video ads. So that's sort of how, I would, think about incorporating your beta feedback into your tool, you know, sort of a product. And then being able to measure that.
[00:42:14] Kabir Bedi: Have we been able to validate, the hypothesis that we really set out, to solve for?
[00:42:19] Andrew Zigler: Yeah. And that's actually another huge parallel with, agentic AI with engineers too, that you just said. Because when you see them start to use it and start to adopt it, they pick up copilot, they're using cursor. it goes like really quickly, overnight, from, oh, they just like maybe queried it once or twice or did an auto complete to, oh, they're using it every day or every time they use their IDE, they're agentically coding or whatever the case you see, in order of a high adoption, right?
[00:42:44] Andrew Zigler: And that's like a strong signal that mirrors what you see with your video generator. Folks go to use it once and then it works and it gets them what they want and so they go use it again. And I can't think of a better result, to ultimately arrive at, 'cause that's how people build products that just like spread word of mouth [00:43:00] and just really solve a problem.
[00:43:01] Kabir Bedi: Exactly, and, and you know, one of the, one of the joyful things about working on AI products like this video generator. is that the underlying tech is improving so rapidly that if you go week over week, month over month, you will see step level changes.
[00:43:19] Andrew Zigler: Yeah.
[00:43:20] Kabir Bedi: that is, important for folks to keep in mind, regardless of the AI application that they're trying, is that not to view it purely from the lens of where it is today, but that the underlying technology powering those, whether it's, you know, sort of coding applications, but in this case content creation applications.
[00:43:37] Kabir Bedi: The, the sort of underlying technology is, you know, sort of rapidly, uh. Innovating and so they'll just keep seeing that progress come out.
[00:43:45] Andrew Zigler: Yeah, it's gonna be like really interesting to continue to follow the product as it evolves, because what you've done is you've laid a foundation, you've created a a one for one transfer of folks coming in with no technical knowledge, wanting to get their really great assets to having the engineers and the [00:44:00] scientists behind it to support that effortless creation on whatever modality they want to work.
[00:44:05] Andrew Zigler: And ultimately, the folks with, the most easy to adopt modalities are going to win. Um, and all of these types of adoption scenarios because when you get a mom pop who uses a tool and it just works and it gets them what they want, I don't think that anything can make something stickier than that.
[00:44:20] Andrew Zigler: So, uh, I, I, I'm really excited to see where it goes, but something I wanted to end our conversation on that I know is top of mind for a lot of our listeners too, is just about the skills evolution that you see for engineers, maybe those on your own team and how it's definitely different than engineers of yesterday.
[00:44:37] Andrew Zigler: Stack overflow generation engineers, the ones that didn't get all of their answers, from their GitHub copilot. So I am curious to know from you, what do you see as things that stand out, as skill sets for the engineers of tomorrow and, and how are y'all building those skills within your own team?
[00:44:54] Kabir Bedi: yeah. You know, for me, what I have been encouraging the team is to rely on [00:45:00] another Amazon. Um. Leadership principle, which is, which is learn and be curious. This space is, prime for all of us to be diving in and learning constantly. And so what I've been encouraging the team to do is heavy, heavy knowledge sharing when it comes to on the application layer or at the underlying model layer. And trying things firsthand as a way to improve, whether
[00:45:28] Andrew Zigler: Mm.
[00:45:29] Kabir Bedi: diving into prompting yourself, seeing what the output looks like, modifying your prompts to see what the output looks like. And so a lot of, you know, sort of my of the team is to really roll up their sleeve dive into the details and,give things a try.
[00:45:46] Kabir Bedi: That would be my advice for anyone in the software engineering science work streams. Working in this space is that you'll really only get better by doing
[00:45:55] Andrew Zigler: Yeah.
[00:45:55] Kabir Bedi: else is also learning with you and share [00:46:00] information with others, learn from others, be in a, in a sort of a growth mindset environment.
[00:46:05] Kabir Bedi: That's really the message that I've been sending to the team and I, and I think it's also brought about, an environment of where folks are quite excited to sort of,learn from each other. Everyone is constantly pinging, this is what I tried, this was the output. You know, and not everything goes well, but that's part of the learning process.
[00:46:24] Andrew Zigler: Yeah.
[00:46:25] Kabir Bedi: and don't, you can't win. By a sort of being on the sidelines, you have to be playing in the field, and so you have to jump in yourself. don't be too harsh on yourself, on the things that you know today. Just have a growth mindset. that would be my only advice.
[00:46:39] Andrew Zigler: Okay. You can't win on the sidelines. I love that. You have to get in, you have to go play on the field. Like what a great way to phrase it. I'm gonna be saying that forever. I am really happy that you're able to kind of paint a picture for us of how this development happened for your team.
[00:46:52] Andrew Zigler: How you're able to build and ship something that has such a huge impact on top of such a rapidly evolving tool. Folks, you know, who, we talk [00:47:00] to and it definitely, this is a common sentiment in a lot of our listeners engineering orgs, like there's a paralysis around even just adopting AI tools and,
[00:47:08] Andrew Zigler: and building them into your workflows, much less turning around and building them, productizing them, taking them to market, and having mass adoption on solving such a huge problem set for folks. So, I think there's a lot that we can learn from your engineering culture, so I appreciate you sharing with it with us.
[00:47:24] Andrew Zigler: for our listeners who are curious to know more about you and and video generator, where would you recommend they go, uh, to learn, about these kinds of things?
[00:47:33] Kabir Bedi: You know, you can follow me on LinkedIn, as well as get a lot more information on. Video generators capabilities, as well as a latest launch on the about Amazon blog, as well as the advertising.amazon.dot com blog. We, we've posted information instead of all of those places as well as, on my LinkedIn
[00:47:55] Andrew Zigler: Okay, fantastic. Then we'll get that shared in the show notes so you know our listeners, [00:48:00] you know where to go to check those out. another call out for our listeners too is that if you like this conversation, made it all this way, that means you did like the conversation. Definitely give it a, like, give it a share.
[00:48:09] Andrew Zigler: most importantly, come let us know on LinkedIn what you thought about it. I'm posting this episode, on our substack newsletter. so if you're not subscribed to there, you can check it out. But it's really easy to join us online. We're both on LinkedIn and we'd love them. Know what you thought about today's discussion topics, maybe even see some of the stuff that you're trying out in video generator if you take it for a spin.
[00:48:30] Andrew Zigler: So, definitely let us know and we'll see you next time on Dev Interrupted.
