AWS Mid-Year Leadership Summit 2025 | Colleen Aubrey, AWS

Clips
News
More from AWS Mid-Year Leadership Summit 2025

Colleen Aubrey

SVP of Applied AI Solutions (S-team)

AWS

AWS believes humans and AI can be workplace allies

Amid artificial intelligence concerns related to job loss and content scraping, Amazon Web Services Inc. has chosen to go full steam ahead on implementing agent workforces and AI-powered tools.AWS believes in the power of AI to transform the workplace, and wants everyone else to buy in too. Otherwise, as tech leaders often suggest, employees who don’t play nice with AI may be left behind.TheCUBE’s John Furrier talks with AWS’ Colleen Aubrey about the company’s AI vision.“We need to be building products that are in the hands of businesses,” said Colleen Aubrey (pictured), senior vice president of applied AI Solutions at AWS. “It’s core to how they operate every day

play_circle_outline John Furrier introduces Colleen Aubrey for a mid-year AWS Halftime Report.

play_circle_outline Integrating AI in Hybrid Workforces: Colleen's Insights from Applied AI Solutions at AWS

play_circle_outline Connect product's AI transformation increases customer engagement and experience.

play_circle_outline Emphasis on knowledge operations within customer-centric businesses and industries.

play_circle_outline Empowering Future Workspaces: Strategic Collaboration Between Humans and AI in Healthcare, Retail, and Decision-Making Moonshots

play_circle_outline Transforming Businesses: Embracing AI Integration to Revolutionize Workflows and Inspire Fundamental 'Why' Questions

play_circle_outline Emphasis on hands-on experience and leveraging AWS infrastructure for innovation.

play_circle_outline Streamlining Business Workflows: The Vision for User-Friendly AI Applications without Complicated Coding

Info
Transcript

Colleen Aubrey, AWS

Colleen Aubrey

SVP of Applied AI Solutions (S-team) AWS

In this AWS Mid‑Year Leadership Summit interview, Colleen Aubrey, senior vice president of Applied AI Solutions at AWS, sits down with theCUBE’s John Furrier to unpack how agentic AI is reshaping enterprise workflows. Aubrey explains why the next generation of AWS Connect, released in March with AI native capabilities across voice and digital channels, creates a launchpad for “knowledge operations” that extend well beyond the contact center. She outlines how usage‑based pricing lets businesses weigh AI labor side by side with human agents, giving leaders a cl... Read more

explore Keep Exploring

What is Colleen Aubrey's role and responsibilities at AWS? add

What insights or reflections were shared about the future of work and the role of technology in business operations? add

What features and capabilities does the Connect product offer after its recent updates? add

What role do knowledge operations play in businesses, and what are some areas of focus for future development? add

What are the key areas of development and innovation being explored in the context of future workspaces? add

What is the current state of workflow processes, particularly in relation to the influence of AI on how tasks are performed and companies operate? add

What are some of the mission parameters that guide Amazon's approach to new opportunities? add

What is being proposed as a solution for business users to manage tasks without needing to use the AWS Console? add

bolt Powered by CUBE AI

Colleen Aubrey, AWS

search

>> Hello, I'm John Furrier with theCUBE are here in Seattle at AWS's headquarters for a special Halftime Report, mid-year review. We're halfway between last re:Invent and re:Invent. A lot of updates. The speed of AI has really been the theme of the year. We want to look back and look forward. Colleen Aubrey, AWS's SVP of Applied AI Solutions, great to see you.

Colleen Aubrey

>>

>> Thanks for coming on. Last time we met we just chatted, we weren't on camera. Great to have you on. Your experience at AWS has a really storied history at Amazon, now at AWS. You built one of the biggest businesses at Amazon and now with this AI Applied Solutions, you are working on some of the biggest ideas at AWS. Take a minute to explain what your job is and what your organization is because I have a lot of questions.

Colleen Aubrey

>> Yeah. So when we met last year, actually, I think I was just maybe month one into this new role at AWS, was trying to work out what the mission was. And so I feel pretty convicted that we need to be building products that are in the hands of businesses. It's core to how they operate every day. It's mission-critical to how they operate. And in this past year, of course, it's become very, very apparent, certainly to me, I'm sure to many, many people, that the future of how we work will involve workforces of agents. It will be hybrid, there'll be people collaborating with AI on a daily business to get the job done. And how that happens is I think pretty exciting area to unpack. So there's the problem of what's the software under the covers, the functionality that we know and love today with a lot of SaaS products and there's a lot of complexity there. But then for me, there's this new layer that sits on top, which is what is the experience of managing a hybrid workforce and a workforce of agents that is your collaborator, your consultant, your assistant? Many different roles that AI can play to actually change the way that we operate our businesses.

>> So is it a business unit within AWS or is it more of moonshots and big ideas? Because I know you got Connect, these large-scale software products that are different than what we see in the marketplace and the ecosystem. Is the scope to take some of these big ideas that have been come out of COVID and look for new business opportunities? What's the focus that you have your eyes on now and medium-term, what's the north star vision?

Colleen Aubrey

>> Yeah, it's a bit of both. So I think the Connect product is in good shape. We released a next generation of Connect in March. It has now got AI native across all channels. So all digital channels across voice, across all experiences, the self-service customer experience, the human agent experience, the supervisor experience, the analytics. So really that's transformed now into a consumer engagement product. And so we think we have a nice place here to work out where I would use the word knowledge operations. It's something we do a lot of at Amazon. I think most businesses have large knowledge operations at the heart of the products and services they're delivering to customers, whether it's loan applications, whether it's managing fraud, whether it's actually managing HR. So I think Connect provides a really nice basis for moving out. And then I have some moonshots, some areas focused around key verticals, retail, healthcare, life sciences that are early in the development. Some work around planning and decisioning, which I think changes now that you can put AI to work there. And then the other piece that are doing a lot of early iteration and prototyping around is what is the workspace that people will use in future? And I don't believe we're going to stay in a world with 25 tabs in a browser and 15 different applications open. I think we will move to a workspace which is much more abstracted from the doing and much more about how and the strategy and the business transformation we're trying to drive, which I think requires a different interface.

>> I love that future of work kind of angle because if you look at post-COVID, there was a big real estate problem. Now let's get back to the office. I think everyone generally agrees getting back to the office is what everyone's kind of moving to or move to, whether it's five days a week or a couple days a week. But you're starting to see people look at their spaces differently post-COVID, post-distributed workforce where it's a lifestyle vibe with technology, you're starting to see new projects even emerge with spaces. So to not make it like, hey, we have foosball and free food, more of this is a great place to work with physical and physical AI, blending it. Physical AI as we see digital twins in factories, all these kind of new capabilities. But the idea of the physical world merging in with digital is a huge trend. That's not a vertical, that's a horizontal play-

Colleen Aubrey

>> That's horizontal. Yeah.

>> What's your vision there? Because that's where things I think come together, I'm just kind of connecting the dots. What's your reaction to that?

Colleen Aubrey

>> I'm not sure. I can't say I have a very crisp vision of where this is going. I just know it's not where it's today. And then I think it will depend on the type of organization. I'm just going to talk about the workplace specifically, I'm not really focused on your personal experience and certainly personal experience with AI. But in the workplace, I think there will be many companies today that are born today or that are growing today that will be fast to adopt a work environment which is hybrid. And they will be looking for how do I form my company in a box in a way where I am able to deploy AI to do specialized work on my behalf and do that in a way where I'm looking at the operation of the whole business? And then I think we have the enterprise customer, there's a different transformation to happen there. There's real questions I think about what is the work that will be critical for people to focus on and how will AI work on their behalf and collaborate them to do that? And that environment I think starts to... The complexity of making that transition I think you start to go from less, how do I whole business operate to how do I think about engaging with my customers? How do I think about managing my back of house, my front of house? And I think we'll see different stages of evolution as the capabilities get better-

>> That's great way thinking about it....

Colleen Aubrey

>> and then as we get transformed-

>> That's a great way of thinking about it. I didn't look at that way. So put the experience aside, that's what my device telling me? Am I feeling good today? That kind of thing too. You're looking at more of the work itself, the workflows. I do a job in the office, how do I be more productive? That kind of piece.

Colleen Aubrey

>> Yeah. I don't think it's this... In a funny way, I don't know if it's funny yet, but in a way everyone will be a manager. Some people will be managing large workforces of agentic AI and some people will be managing both people and AI. But in that case, my role changes from the doing to the orchestration to the, what's the problem to be solved? What's the blocking and tackling? I do think that AI, you'll have agents come back and say, well, I think this is the answer, another agent will say, well I think this is the answer and we'll have to arbitrate and we'll have to make decisions actually we're going this way and point everyone in the right direction.

>> Matt Garman was mentioning on my interview with him this morning that we're waiting for the ideas. Ideas are going to be the scarce resource, that's going to be the value, the human in the loop. Obviously there's a lot of talk about agents replacing jobs. I'm very much on the record saying I've seen that move before. ATMs are supposed to kill the bank teller. There's more bank tellers than ever before. We've seen this. There certainly will be replacements. The human driving the AI will be key. What's your reaction to that? And as you look at the agentic workforce words are coming up like the word labor. It feels like a spreadsheet item like labor costs or digital workforce, digital labor, digital workforce, digital worker. I mean an agent can be a digital worker but do half the task and then the human... So how do you think about these words and how people are framing the conversation around digital agents as a function of getting work done?

Colleen Aubrey

>> Yeah. Now keeping in mind my group we've chosen specifically the name Applied AI because that last mile work of putting AI to work in your business and getting the results you're looking for, I think that's very hard. Your customers are not going to tolerate a 20% error rate. There's just going to be... Not feasible. And so I think there's a lot of work to do on that last mile to have AI working in the way that you want it, delivering the results you're looking for at the quality with the constraints that you have in your business. And so, I think this takes a while to evolve. With Connect I think we have a very nice starting point. Solving for customer interaction and customer service I think is an area where we've been applying ML for a long, long time, and so the application of AI gets there. And so if you think about the self-service chat experience or even the self-service LLM-powered interactive voice. Today, what we've adopted is a usage-based pricing. So we're saying that when AI interacts on your behalf on voice, then you should pay for it as a voice service. The same as if you connect a call to a human in the contact center, then you pay on a usage page for a voice service. So our interpretation here is that there's value in the work that's being done and if you deploy it well it should be equally as valuable as actually the next phase, which would be there's a problem that's more nuanced and I need a human to get involved here. And so we are very much taking the point of view of like for AWS we are starting with usage. I don't know what the future is. I can imagine contracting AI by the hour. I need a legal contract, AI, I want you to review this contract and come back with feedback. How long do you want to do this? I need three hours. I'm going to pay you this much for three hours. I can imagine that being the case. Which is more getting into the idea of this is a labor force and that's how we might purchase it. But it'll be interesting to see how that goes.

>> It's interesting and the point about SLA is kicking in, if I'm going to pay it better be good. It can't be wrong, to your point.

Colleen Aubrey

>> It can't be wrong.

>> So we don't know what we don't know yet, is essentially kind of where we're at. Well, we do know that we have a lot of data and you have experience with Connect. There's a treasure trove of experience you probably have gleaned out of that that's helping you apply some of the thinking around business opportunities, potential areas to optimize, where you want to take the initiative or not. What's the learnings you have now at your level from Connect, your previous experience at Amazon, building up that business with new models, whether they're new business models or new AI models come new experiences, new patterns, so new user patterns, so there's things are changing, what are some of the things you're seeing there around what you're going after? How do you make the selection what to work on? What have you learned from the... I guess my question is what do you know? What do you know and what do you don't know? And then how do you use that information? Because a lot of people I hear are saying we're using AI to figure out what we have and then figuring out how to optimize it. And some of the projects, even Matt Garman was saying, Strands came out of something that wasn't intended. So you have these kind of good consequential benefits of playing around.

Colleen Aubrey

>> So I think for me with Connect, there's the problem of getting the performance that you need in your business today and you can't tolerate high defect rates there. But I think once you get confidence in the AI's ability to act on your behalf in a way that you feel good about, the question for me is now, well, how would you transform how this gets done? And I think some of the areas that we're looking at, for Connect some of our customers are doing really interesting things where they're actually using Connect to manage things like back of house processes. So let's take this out of the contact center. Even within Amazon, our HR team uses Connect to manage interaction with employees and to manage benefits and all sorts of employee engagement. So if we look at that, then the first job to do might be like, well, let me just reduce toil and let me look... Automation, people use the word automation-

>> Mundane stuff.

Colleen Aubrey

>> That's right.

>> The toil.

Colleen Aubrey

>> That's right. But the real question for me is, well, once you have confidence that AI can do the right thing, then how would you transform this workflow? And I have conversations with my team at the moment exploring new product areas. And we review a PRFQ, a working backwards document, I'm like, okay, this is about automating how it happens today, but with the capabilities we have then how should it happen in future? And we've seen this happen again and again in technology. Where you actually have these breaking changes in how the experience happens. And I think this is the opportunity for organizations and the opportunity, certainly for my group to bring a point of view on how can we put AI to work for you in a way that reinvents how businesses get done?

>> Yeah. One of the things we talked about last time we met and I wrote about it on SiliconANGLE was you're kind of in a good position to leverage and tap the multitude of benefits that AWS as a infrastructure has. It's like a sandbox. It's almost an incubator for you at some level. So there's a huge aperture of things to play with to look at big ideas. You mentioned moonshots like healthcare, we're seeing a major resurgence and say healthcare and life sciences because, one, they documented stuff in regulated environments and, two, they've never really been great at IT. I'm saying that with a little bit of caveat, but for the most part they weren't like a higher end IT. But AI helps them so much. And so we're seeing a massive resurgence there. Life sciences, they're getting access to supercomputing capability that they never could get before to run the compute they need to run, the complexity of that stuff. So everyone's sitting there saying, wow, what a great opportunity. So it seems like these new markets have new characteristics. How do you view that and how do you go in there from an Amazon perspective, what are you tapping from AWS to learn and say, okay, we're going to enter healthcare at this layer, large scale, this could be billions of dollars worth of business, but yet support the ecosystem?

Colleen Aubrey

>> Yeah. So I mean, I'm Swami's first best customer. Everything that he's building to enable a agentic applications, this is where I'm his on his door saying, hey, this is what I need to be able to build applications that will work every day for a business. And so I think from that perspective, Swami's sort of my partner in crime and someone that I'll be really-

>> He's your best friend.

Colleen Aubrey

>> He's my new best friend. Whether he knows that or not, he's my new best friend. And then of course I get all of the value that external customers also get from AWS and the capabilities there. I also have the experience of Amazon. And so in healthcare we're working very closely with the Amazon health Services Group and Neil Lindsay, We have One Medical, we have the pharmacy business there, they're both users of Connect for their customer experience. And then with One Medical, we are using HealthScribe today to do ambient listening for doctors in One Medical to be talking with their patients, the ambient listening is taking their notes. Now the question is can we translate into that into prescriptions that get sent out, referrals that get sent out? Can we reduce that toil in the patient-doctor experience to increase the amount of time they're spending face-to-face and decrease some of that work that is necessary but lower value?So for me in healthcare, it's interesting because I have a great partner inside of Amazon in addition to the capabilities of AWS to be able to -

>> So Neil runs that group?

Colleen Aubrey

>> That's right

>> Now, what's your relationship with that group?

Colleen Aubrey

>> Well, we are working very closely to develop the capabilities.

>> And so you're going to build a suite on top of it or Pan-Amazonian?

Colleen Aubrey

>> So I think of it as coming from two directions. One is from, we have a lot of a good number of healthcare customers using Connect. And so they're already pushing out to more of their patient experience. And so there is work that we can do to make that easier for them, to make that more relevant in healthcare. And then working with Neil, there's work from the doctor, the provider point of view that goes back and actually connects that customer journey. So I'm working from both sides.

>> It's interesting. What I love about your background is because you were on Amazon on the retail and obviously the ad side, you built that business up from nothing to huge success. I was in Hawaii, I was mentioning before you came on camera, my trip last weekend, one of the people pulled out their phone and I saw the One Medical app and I said, how do you like that? They love it. Because it's healthcare, it's like Amazon getting packages. We know the benefits of-

Colleen Aubrey

>> It's very convenient....

>> faster, fast, lower cost, high quality. And that's the ethos of Jeff Bezos' original vision. It makes so sense to have that be for healthcare, as a user. Why wouldn't I want to just find the best doctor that knows me? And with agents, I just see a massive opportunity to make everything work better. It's a huge holistic view. It's a lot of moving parts because you have the provider side just like keeping notes, who keeps notes? That could affect all kinds of billing issues, referrals, slows down care. On the user side, it's like I got to wait for my scan. These are so antiquated workflows from a time perspective.

Colleen Aubrey

>> And I think AI can... A lot of the workflows I think have been built up over time, there's regulatory requirements that need to be met. And then there's just the complexity of running large ambiguous operations every day that are mission-critical for so many people. And I think there are many opportunities for us to reduce that toil, improve the customer experience. We're all about customer experience. Connect is all about customer experience. So I think we have a really nice base of capabilities there with the AI that is customized to healthcare, to actually put patients more in control of their healthcare and actually to enable providers and many of the knowledge workers that operate in the background there to actually be much more-

>> So your lens is wide, you have a wide lens looking at a holistic picture. I brought up with Matt's interview, I was fascinated by Andy Jassy's shareholder letter this year, where he asked the why culture, the why generation. So I want to ask you the why question. And there's a favorite meme out there of the famous Steve Jobs when he was young, he was talking about Apple computer. He had this famous meme, it's like when I joined Apple I asked, how do we know what it costs? Well, we used mechanism and we round out things. Because they didn't know what the cost was. So he's like, why? So he built a factory that knew the unit cost of everything. So the point was is that he just asked why and the answer from the conventional wisdom was that's how we do it. When they actually could get to the why. So a lot of people are asking themselves, why are we doing things this way? So I think we're in a workflow disruption mode. Big time.

Colleen Aubrey

>> Yeah. I think so. I think AI is this incredible catalyst for asking the why questions of why do we do it this way and why wouldn't we do it in a different way with the capabilities that we have in our hands today? And for me and my group, that is the question. Why do we do it this way? Why do companies do it this way? And what have we learned within Amazon? We've been doing hands off the wheel work for many, many years of how do we actually put machines to work on behalf of people and collaborating with people to be able to scale and operate efficiently. There's a lot that we've learned ourselves. Why can't we build that into products that actually help many customers?

>> So you're like the incubator, agent, tie all the Amazon goodness together, AWS goodness together, looking at the big opportunities. Is there a mandate? Do you have a threshold? Some corp departments say, we only look at billion dollar transactions. Is there a scope to your focus? You say, I'm looking at only billion dollar opportunities or $10 billion opportunities. Share some of the mission parameters.

Colleen Aubrey

>> Yeah, I think it's sort of consistent with how Amazon has approach this for many years. It's like how big is the opportunity? Is it really a sizable opportunity, a sizable playing field? Is it right for disruption? Is there something that we can do that's uniquely different than what's happening today? And if we can get answers that are yes to those questions, now it's execution. And you've been working with Amazon and AWS for a while, execution is something that's in our wheelhouse. For me, the real challenge today is getting really crisp on how can we actually deliver something that's meaningfully better than what companies are using today? And do we really have a right to do that? Is this something that we understand some capability we have that gives us a unique point of view that we can do that? I think Connect is an interesting place that we started. We have one of the largest customer service teams in the world with our stores business, with our AWS support business, with our seller business. We have many teams in Amazon using it. The healthcare businesses use Connect, so that gives us a unique point of view. But we have many teams doing knowledge work at high volumes that I see as a point of inspiration. And then we have these examples like healthcare and life sciences, which I think the capabilities that we can put in the hands of those customers today have never been available before. And really, it's compute, it's the use of AI, it's the use of technology to make things happen in a faster, more effective way than they have in the past. And those things give us, I think, the opportunity to innovate in those areas.

>> Colleen, what's the top three areas on the execution side that you're focused on? What are the top three? If you have to stack rank them.

Colleen Aubrey

>> Yeah, we have a moment in time with Connect that we have to take advantage of, and we have an AI native product that we launched in March. I can't imagine a better timing for an inflection point for that product. So number one, we have to be very good at landing that and demonstrating to customers the value. Number two, the interesting thing for me is that we also have to transform the way that we work. If we are not AI native in the way we're building and operating every day ourselves, who are we to be able to envisage what that looks like for others? So my number two is getting everyone in my group hands-on using AI every day and transforming how we work because by doing that we are going to learn a lot. And then number three is a little bit, I guess consistent with Andy's focus is that speed matters and more than ever, speed matters now, and it also means that we have to operate like a startup. So I'm constantly sort of saying, do you really need my approval on this? Why? Why can't you make this decision yourself? I think you have everything you need to be able to go. So let's not slow down. Please go faster. Lean into prototyping. AI tools today can help us prototype so quickly. And I think this is maybe an area that we had lost a little bit, the art of. Things like that I think are my number three.

>> Every time I talk to you I want to work for you. It's like, what a great assignment. You must get that a lot because I mean, think about all the ideas of just I have, what I would do with AWS. I have a bias for action, one of the principles of AWS and Amazon. There must be a lot of people who have ideas about how to make the most out of AWS and Amazon's infrastructures assets. I mean Connect's a great example. There's so much you could rally around Connect. You got data, you got touch points, so there's a lot of tentacles in organizations. That's a great place to start. What other ideas are coming out from your team, from people probably wanting to join your team because it's kind of a playground?

Colleen Aubrey

>> It is a bit of a playground, I-

>> It's a fun playground by the way. Business playground.

Colleen Aubrey

>> I sent an email to my group at the beginning of the year in the email I was talking about what's this year going to look like? What would I like to say in December? And one of the things I put in there was that 2025 is the year of the PRFAQ. It's the year of getting new ideas out there and just rapidly iterating on what those ideas might be. So we are several dozen through and there are a small portion that really pop, we have something here. And so we are-

>> PRFAQs, which is the working backwards document that come in from the team?

Colleen Aubrey

>> From the team. Yeah. So it's the year of the PRFAQ. The question for me is really back to what we were talking about before, is it a huge opportunity? Is the area ripe for disruption and can we bring something, a different point of view and a unique capability that really delivers something meaningfully better? And the answer to those things is not always yes. But I really like that people are really pushing the boundaries and are excited, frankly, about the opportunity to really break open the canvas.

>> A couple of things I want to get your thoughts on because I think they're kind of trends that are perfectly aligned with the wave that we're on. One is the old expression, do more with less. We've heard that cliché in companies when they're downsizing or budget reductions or want to be profitable. I know Andy has publicly said this and I've heard some internal emails have gone to you, even said you've sent an email to your team to use AI, not to reduce costs, but to get muscle. So that's one thing. Do more with less because AI can help you do more with less, be more productive. It's one of the benefits of AI. The other thread is that when you do AI projects, you tend to learn other things. And Matt Garman gave the example of Strands, which I said earlier. That was just testbed basically to solve a problem. Well, this could be good for everybody. It's now a PRFAQ and it went out the door. Hugely successful. So there's a lot of by accident on purpose, things happening. That's a huge factor. So both those things are in play. People are going to have to get the muscle to use AI, this is a great recommendation. Talk about those dynamics because that means it's going to be faster movement. People are going to have agency and authority, which is not always on everyone's management list of saying, I'll give you enough rope, we'll see, maybe run a little bit. But how do you unleash innovation because you're going to have chaos on your hands, too many ideas?

Colleen Aubrey

>> And the first one on do more than less. I don't know if it's do more with less, really. For me, this is a case of, you know, our technical capacity, let's say the capacity of the engineers in my group, that might not be the constraint anymore. So it's not necessarily do more with less. It really then comes down to-

>> Do more with more.

Colleen Aubrey

>> Do more with more. And that I think is the opportunity, doing more with more. And if you don't have the constraint of the number of engineers in the group or the number of applied scientists in the group, but the constraint is the quality of the ideas, the quality of the thinking and the quality of the execution. I think that changes the conversation. And this is a changing conversation I'm super excited to embrace. And then the next bit is, for me, one thing I've learned working in Amazon for almost 20 years now, you have to embrace the chaos. Trying to control it I think that's-

>> Let chaos reign. Reign in the chaos when it's working.

Colleen Aubrey

>> That's right. And for me as a leader it's important to share what's the vision that we have to rally around? What are the principles and tenets we're going to operate with so we can make good decisions? What are some of the mechanisms we use to stay connected and when do we have one-way doors? If people are racing through two-way doors and learning quickly, that's awesome. But when we hit those one-way doors are we really clear on what they are and we actually get in a room and we have a good conversation and we can march forward?

>> One-way doors are Amazon's term for decision that you can't walk back. Two-way doors you can make a test and kind walk it back. Which is actually a good thing. It's a feature, not a bug. The one-way doors are bad.

Colleen Aubrey

>> One-way doors just require a pause.

>> You're committed. You're committed.

Colleen Aubrey

>> Yeah.

>> It's like building a data center.

Colleen Aubrey

>> Yeah.

>> You're stuck with it until you reboot it. All right. So this idea of building muscle, I kind use the analogy of the horse and buggy and cars. You're saying that to your team, learn AI to learn what more is, to do more with more is the mission.

Colleen Aubrey

>> Yeah.

>> How's that going? Because I can see that really being kind of more of a learning curve. So you're basically saying learn AI as fast as you can because it's a superpower, not replacing anything.

Colleen Aubrey

>> That's right. The earliest group I guess to benefit from AI as software engineers. And so with this group, I'm seeing a really nice curve of actually daily use of our engineering tools, the Deepak and team are building. So I'm pleased with the curve. And then what I'm really pressing my principal engineers on and senior principal engineers is like, where are the stories? I want them sharing stories with the rest of the group of something that they have done that has taken it from a two week effort to a one day effort. And those stories are starting to emerge and it's every-

>> Can you share a story? I mean stories drive movements, you know that.

Colleen Aubrey

>> There's one applied scientist, senior principal applied scientist, he works on the Just Walk Out business doing some fine-tuning and models. Had a whole bunch of images that he wanted to get feedback on. And this is something where you get really, it's like the team getting close to customers. It's like, let me just get this out. I've got X number of images. You're going to get 100. I'm going to get 100. Annotate them and I'll use that for fine-tuning. Historically would go to a SaaS product to do a survey, starts digging into that. It's like, wow, there's a lot of hoops to jump through. There's a lot of processes, there's constraints, I have to define 1,000 things, but he really just wants to do something relatively scrappy. Get it done now. He used a coding assistant, used a Q Developer, writes an app for this survey, which served his purpose immediately and was able to deploy it in a day. So he did a day of researching, how do I actually use existing survey tools? And then was like, this seems like way too much work. Spent half a day writing his own tool using Q Dev, and then was able to deploy it and get the results that same week. That would've been something that normally would take him two to three weeks to accomplish, and he was able to do it in two days.

>> It's interesting, I was talking with Dave Vellante on our theCUBE and we were debating talking about the productivity gains with AI and all that, blah, blah, blah, all that stuff. We said we compared the productivity suite of say Excel back on, I'm old enough to remember Lotus 1-2-3 ran Excel, but-

Colleen Aubrey

>> I used Lotus 1-2-3.

>> So it was like, if you knew how to do the macros and if you knew how to drive Excel, you rose to the top.

Colleen Aubrey

>> Yeah, was awesome.

>> You were as awesome. You're like a spreadsheet king and queen. Okay, so now the AI is similar. If you know how to drive that AI, you can do more. But the people who had the Excel mechanics, were usually more technical finance-oriented, not the business user. And we're seeing AI from young people not in computer science wanting to learn how to code, not to be a coder, but to use AI. So there's now a new generation of business AI coders coming out. What's your view on that? Because that example really is telling because I see everyone under the age of 30 basically saying, I'm going to master this AI. So the question is, do I have to code for it? Now Swami just said, oh, just log onto the console. Well, they don't really want to deal with the console. Is there a faster way?

Colleen Aubrey

>> Yeah, my focus is on the business user, and so I will be looking to put all of those capabilities of being able to say, oh, I have a problem to solve, being able to describe that in a natural language, to be able to test that in a natural language, to be able to see an outcome. And say, okay, agent I've just built an agent effectively, go, do this. Come back, check in with me. I want to audit the results. Let me make some adjustments. Now I trust you. Please go and execute that. I think that's very reasonable for a business user to be able to do without going into the AWS Console at all. And to have an environment, which is an AI-powered environment is what we're really working on envisaging, an AI-powered interface where the AI is like my personal assistant in my work that works across many other capabilities that are under the covers to enable me to get beyond these constraints in the past where I would've had to engage any number of different people in the organization to be able to get a task done. It just seems a very real thing that we can be doing today, and that will be -

>> I think that's really real, and I think that's the power way we're seeing is that this is the new generation of builders.

Colleen Aubrey

>> That's right.

>> And the console's great for DevOps and managing LAMP functions and serverless. But this is a great example, I was talking to a young person I was mentoring in healthcare, and she and I were talking and she's like, I'm in a job where I have all this process, clinical trials, ins and out, operations person knows everything, has domain expertise and is looking for work and wants to get into the AI side of it and she asked me my advice. I'm like, well, mm. She goes, should I code Python? I'm like, well, that's yesterday's answer. I go, well, you know all the workflows.

Colleen Aubrey

>> Yeah, that's right.

>> Why don't you just identify the areas that you could just magically with a magic wand-

Colleen Aubrey

>> The clinical trials, there's so much to do there.

>> What would be your advice to that person? Because they're new and fresh and smart, but would be perfect to inject and apply their knowledge with an AI assistant, but they don't even know how to... What do I do? How do I start? That seems to be the question. That's a common theme I'm hearing across all verticals.

Colleen Aubrey

>> I think that knowing the ins and outs of how the business operates, the complexity of how the business operates, the ways in which the customer experience breaks. For example, in this case, with this person you're speaking to, they understand the ins and outs of how clinical trials operate and this is not easy. This is a very complicated and long process today. That's, I think-

>> That's intellectual property.

Colleen Aubrey

>> That's intellectual property. And so transforming how that happens using AI, I think, just gets easier and easier because the deploying of technology to change that will be in her hands. And so for me, I'm like-

>> She just wants to get a job. So there's no job opening. So I'm like, okay, start a blog, get it out there, document it all, and the world will come to your doorstep. So again, this is just-

Colleen Aubrey

>> If I'm here I would be looking for... There's a really, I think, pretty vibrant community of startups in life sciences and in the healthcare that see these opportunities. I would be going there because her real world experience will be invaluable. You can't reinvent what happens today if you're naive about how that really works and you've got to put your hands on every step.

>> And you can drive the AI. It's like the new roads or paving, Amazon Web Services and all the physical infrastructure from Just Walk Out to Connect. These things are now input into driving the AI, in this case, applying it.

Colleen Aubrey

>> Yep.

>> Okay. Your goals for the rest half of the year, you got re:Invent in six months, kind of doing a halftime report. Looking back since re:Invent to now and going forward, what's happened in the business and then what do you see in the second half of the year and what are you going to be driving into 2026?

Colleen Aubrey

>> Second half of the year we have some, I think, exciting new products that are in early stages. Some with our first customers, with their hands on them, some that rapidly iterating through prototyping. I would like to see a few of those land for re:Invent with some good customer feedback on how that's changing their business. So that's a near-term goal for me. Moving into 2026, when I joined AWS and started this role I sort of had a grab bag of things that have been sorting through. Now I think I have a portfolio. By 2026 I want to have a family. A family of agentic applications that come together for businesses and a user interface that helps them move into this future of managing these hybrid workplaces.

>> Colleen, you're an inspiration. I'd love your job. I'm looking forward to continuing to, one, keep in touch, but also your view is a holistic long game view of where we are today and kind of where we're going, it's a great example for what's coming, certainly for the ecosystem and certainly for AWS. Thanks for coming on.

Colleen Aubrey

>> Thank you, John.

>> Okay, we are here for the Halftime Report. I'm John for theCUBE. Thanks for watching.