Skip to content
Lenny's PodcastLenny's Podcast

Asha Sharma: Why org charts give way to agent work charts

How products become living, learning organisms with the loop at the center; Sharma on post-training, reward models, and work charts replacing org charts.

Lenny RachitskyhostAsha Sharmaguest
Aug 28, 202557mWatch on YouTube ↗

EVERY SPOKEN WORD

  1. 0:00 – 4:18

    Introduction to Asha Sharma

    1. LR

      You said that we're just starting to scratch the surface of what an agentic society actually looks like.

    2. AS

      We're approaching this world in which the marginal cost of the good output is approaching zero. We're going to see exponential demand for productivity and outputs. The way that you scale to that is with agents. When all of that happens, the org chart starts to become the work chart. You just don't need as many layers.

    3. LR

      We were chatting about this concept you have that we're moving from product as artifact to product as organism.

    4. AS

      Because these models are so effective at this point, you want to start to tune them to certain types of outcomes. All of a sudden, these are these living organisms that just get better with the more interactions that happen, and I think this is the new IP of every single company, products that think and live and learn.

    5. LR

      Planning right now is just crazy. How does anyone plan a roadmap when there's just like, okay, GPT-5's out?

    6. AS

      We think about it as, what season are we in? Season one might have been prototyping of AI, and then it was all around models and reasoning models, and now, it's the advent of agents.

    7. LR

      (instrumental music) Today, my guest is Asha Sharma. Asha's chief vice president of product for Microsoft AI Platform, where she oversees their AI infrastructure, foundation models, and agent tool chains, while also leading applied engineering, responsible AI, and growth for the core AI division. She was previously COO at Instacart and VP of product at Meta, where she ran Messenger, Instagram Direct, Messenger Kids, and Remote Presence. She also sits on the boards of The Home Depot and Coupang, and she's a second-degree black belt in TaeKwonDo. Asha has a really unique and rare role that allows her to see more than most anyone else in the world where things are heading with AI and what works and doesn't work for companies that are building large-scale AI products. In our conversation, Asha shares a bunch of trends and predictions that she's seeing that I haven't heard anyone else talk about, why we're moving from a product as artifact to product as organism world, why GUIs are being replaced by code-native interfaces, why post-training is the new pre-training, the coming agentic society, what it takes to be a successful builder today and going forward, and also her single biggest leadership lesson that she learned from Satya, who she works closely with. If you enjoy this podcast, don't forget to subscribe and follow it in your favorite podcasting app or YouTube. Also, if you become an annual subscriber of my newsletter, you get a year free of 15 incredible products, including Lovable, Replit, Bolt, n8n, Linear, Superhuman, Descript, Whisperflow, Gamma, Perplexity, Warp, Granola, Magic Patterns, Raycast, ChatBRD, and Mobbin. Check it out at lennysnewsletter.com and click Product Pass. With that, I bring you Asha Sharma. This episode is brought to you by Interpret. Interpret is a customer intelligence platform used by leading CX and product orgs like Canva, Notion, Perplexity, Strava, Hinge, and Linear to leverage the voice of the customer and build best-in-class products. Interpret unifies all customer conversations in real time, from Gong recordings to Zendesk tickets to Twitter threads, and makes it available for your team for analysis and for action. What makes Interpret unique is its ability to build and update a customer-specific knowledge graph that provides the most granular and accurate categorization of all customer feedback and connects that customer feedback to critical metrics like revenue and CSAT. If modernizing your voice of customer program to a generational upgrade is a 2025 priority like customer-centric industry leaders like Canva, Notion, Perplexity, and Linear, reach out to the team at interpret.com/lenny. That's E-N-T-E-R-P-R-E-T.com/lenny. Today's episode is brought to you by Dx, the developer intelligence platform designed by leading researchers. To thrive in the AI era, organizations need to adapt quickly, but many organization leaders struggle to answer pressing questions like, which tools are working? How are they being used? What's actually driving value? Dx provides the data and insights that leaders need to navigate this shift. With Dx, companies like Dropbox, Booking.com, Adyen, and Intercom get a deep understanding of how AI is providing value to their developers and what impact AI is having on engineering productivity. To learn more, visit Dx's website at getdx.com/lenny. That's getdx.com/lenny.

  2. 4:18 – 6:20

    From “product as artifact” to “product as organism”

    1. LR

      Asha, thank you so much for being here, and welcome to the podcast.

    2. AS

      Thanks for having me.

    3. LR

      I want to start with something that, uh, we were chatting about before this that I've never heard about as a concept that I think is, uh, is gonna be really helpful for people to think about, which is this concept you have that we're moving from product as artifact to product as organism. Talk about what that means and what people need to understand here.

    4. AS

      It's a, it's been a pretty interesting shift, especially over the last year or so, because when I got to Microsoft, um, it was kind of right after, uh, OpenAI and the large foundation models happened, and then immediately after, there was this explosion of models, uh, proprietary, open, uh, frontier models that were pushing the frontier curve. And so, they were both, uh, more efficient, and then we're starting to see domain-level expertise in a bunch of them. And then, you know, even more recently, models now can, you know, tool call and they can function call and they can take action. And I think that's just giving way to a new type of products that are, uh, starting to see, uh, some success. And so, all of a sudden, products aren't just like these static artifacts that we start to ship. It's not just like, hey, come up with an idea or an insight, go solve a problem, ship it into the world, maybe make it a little bit better, and then have a dashboard. All of a sudden, the, the whole KPI is, what is the, the metabolism of a product team to be able to ingest data and then digest the rewards model and then create some sort of outcome? Because these models are so effective at this point, you want to start to tune them to certain types of outcomes, whether it's price or performance, quality. And so, it's pretty exciting because, all of a sudden, these are these living organisms that just get better with the more interactions that happen, and in many ways, I think this is the new IP of every single company, and it's a completely different way to build product and to even think about, you know, products that think and live and learn,

  3. 6:20 – 9:10

    The rise of post-training and the future of AI product development

    1. AS

      which is kind of exciting.

    2. LR

      So when I hear this, what I'm thinking about is when I had, uh, Michael Tirrell on the podcast, the Cursor CEO. He talked a lot about how their big moat is the data that they capture from people using Cursor, selecting- accepting certain suggestions, not accepting other suggestions. Is that what you're talking about here? Just, like, the proprietary data that companies gather from people using their product? Or is there something beyond that even?

    3. AS

      I think why we're seeing, like, the rise of post-training happen is just that the- the models themselves, like, are so powerful. As of this year, Nathan Lambert, uh, did this study that I thought was pretty interesting of all the top leaderboards, and it showed that, you know, once a model hits 30 billion parameters, the CapEx to actually train a model and put, you know, billions of tokens into a, uh, kind of pre-run kind of doesn't economically make sense. Uh, and you can kind of start to optimize on the loop. And so, yeah, in many ways, I think you can... I think using your own data is the best way to do that, but you can synthetically generate data. You have to come up with a rewards design. You have to actually roll it out. You have to A/B test it rigorously. You have to find the job to be done or the use case that it makes the most sense for. And then, yes, like, that generates data that you can learn, uh, from. I haven't ever seen it be one loop, um, for any- any sort of product. I- I think it's multiple tracks running in- in parallel that are kind of like assembly lines, if you will, um, and kind of producing that.

    4. LR

      And so is this kind of, uh, thesis that we're moving towards product as organism, is this- is this basically for model companies or is this also true for, I don't know, SaaS businesses and tools, end-user tools?

    5. AS

      Look, like, I think that, um, software as a primitive is changing and kind of the artifact inside of it, uh, is- is m- a model alongside the software components itself. And so, in many ways, I think that, you know, software products will all be model forward products, if you will.

    6. LR

      This reminds me, when I just had, uh, Nick Turley on the podcast, who we were talking about before we started recording, head of ChatGPT, and, uh, I was asking just, like, how much does ChatGPT change with GPT-5 coming out? And he's just like, "It's the same thing. They're the same product." It's just, like, G- the- the model tells us what to do in the product of ChatGPT. And, um... And it makes me think about something else of just, like, you would think, why can't just GPT-5 build its own user interface? Just, like, as you use it, it just evolve. It's sort of what it's doing with Canvas and art- and all these things. But, like, that's, like, another way I think about when you talk about this idea of product as organism, is the product... The UX can shift based on how you're using it and evolve automatically without having the product teams have to do anything.

    7. AS

      I 100% believe that's where the world is going, and that my experience should look and feel different than yours. I mean, that's kind of been the advent in personalization, but now you can do it on the fly in the future. So, I think that'll be a pretty fun world. Uh, I also think it will look different for agents and it will look different for kind of power

  4. 9:10 – 12:01

    Successful AI companies: patterns and pitfalls

    1. AS

      users and new users and all of those things, too.

    2. LR

      Let me, let me kind of zoom out a little bit and ask you this question. You work with a bunch of companies that are building AI products on your platform, other platforms. Imagine some just do an awesome job and are killing it. Some are struggling. What do you find are kind of common patterns across the companies that do really well and have a lot of success building really successful AI products, and ones that don't?

    3. AS

      Yeah. So, uh, I think there's things that are kind of more broadly applying to the organization themselves, and then there's, uh, things that are applying to the people who are building the- the AI, uh, products too. So, more broadly, I think there's- there's a pattern that's starting to emerge for successful companies. Like, one is they are embracing AI and everybody becomes AI fluent. So, I think everybody's using some sort of copilot or some sort of AI in their day-to-day workflows. Like, job one, so everyone's not afraid of it, understands how we can raise the ceiling, and- and kind of lower, lower the floor for, like, all sorts of skills and- and- and tasks. Number two, from there they start to say, okay, how can I take a process that already exists and apply AI to making it better? That might be, uh, something like customer support or taking fraud down from 15 days to kind of cure to- to 10 days, and, like, going through that entire loop of mapping out the process, applying AI to it, seeing some sort of impact, and then feeling the- the P&L or the kind of intrinsic benefits that that- that looks like. The third thing then is, like, okay, great, now that you've seen impact, you're- everybody is using it. How do you actually use it to inflect growth? And that can be something like improving the customer experience so your LTV or retention improves. It could be co-creating a new kind of set of concepts or categories. It could be, you know, going from agents that are embedded to agents that are embodied and then being able to take on, you know, exponential number of tasks. I think that where companies fail is that they're doing AI for AI's sake. They have a ton of projects that they're kicking off at the same time without a blueprint to understand how it actually work from what their stack looks like, and they are- aren't treating it like a real investment, and so they don't have the measurement and the observability and the evals all kind of set up. It's gonna do that end-to-end. I think the tricky thing is for enterprises is the- the technology is changing. There are something like 70,000 enterprise tools, like, in the AI space launched last year. It's really hard to know which one you should use for what outcome. And so you really need to bet on a platform or some sort of app server type layer that allows you to swap things in and out and not really be beholden to anything, any- any one technology or any one tool, because the reality is, is the whole thing is going to change. Feel like you have to actually build for the slope instead of the snapshot of where you are. So that's- that's kind of what I see at the

  5. 12:01 – 14:15

    The evolution of full-stack builders

    1. AS

      enterprise level. I think the- the builders themselves are actually changing pretty fundamentally too, right? Every single advent, like change in technology has invented, like, a- a changing set of roles. Um, like mainframes to PCs, like the whole garage engineers, and then when we went from, you know, server to cloud and mobile, there was, like, SEO specialists and CDNs and, you know, growth PMs and UXR and- and- and, you know, front end, back end, and yada yada. And now I think we're seeing this advent of- of the polymath and where, um...... I, I think that full stack builders are kind of having their renaissance, where if you take, like, an average organization, it takes probably 10 steps to launch a product. Um, it could be security review, it could be spec, it could be, you know, user research, and there's what? Five plus functions? Um, maybe s- maybe six or seven, I'm being generous, for a- for a normal organization, and then you have, like, six or seven layers. So, all of a sudden you have 500 different touchpoints that have to happen to get a product out, and when there are 500 models available a week, or 500 new technologies, that is just insufficient. And so, I really believe in- in the concept of a full stack builder. You're seeing it with a bunch of the AI native companies that are coming up. I'm even seeing it in enterprises that have been around for 50 years starting to operate in that way, and I think that gives you velocity and throughput, and then gives you the whole loop to start to actually, uh, metabolize and go through that much faster.

    2. LR

      That's definitely a recurring theme on- on, in these conversations is just kind of the Venn diagrams of PM engineering, design are starting to converge and more and more of other disciplines within your role, so PM needs to level up on design and/or engineering.

    3. AS

      Yeah, I completely agree. I think it's all about the loop, not- not the lane here, um, and so I think that whatever, uh, function you are, you have to be obsessed with trying to understand, like, the efficiency or the cost of- of the product, the actual rewards, or y- like, you know, system design that you're going after, the actual UI/UX, how that actually manifests for agents or people. You have to start to get really

  6. 14:15 – 16:24

    “The loop, not the lane”—the new organizing principle

    1. AS

      good at that really quickly.

    2. LR

      I like this phrase you just used, the- the loop and not the lane. Can you say more about that?

    3. AS

      Oh, it's just going back to our- our previous discussion on, you know, the signals loop and, uh, products evolving and becoming these living organisms and not these artifacts, and if- if you think about getting really good at that loop, I think that is the product, that is the IP, that is the future of every organization, and I think feedback becomes continuous and observability becomes the culture, and I think that, um, functions start to- to blur, um, in future workforces.

    4. LR

      To make this even more real, is there an example of a product or a company that is a really good example of this, of d- doing this well, living this kind of loop life?

    5. AS

      I think most companies that we're seeing in the space from an AI perspective are doing this. I can tell you about a couple that we're working on. Obviously, in- in the coding space, you mentioned Cursor. GitHub has very similar features that we're using, kind of as an ensemble of models that have been fine-tuned across, you know, uh, 30 different countries, all of the languages, to actually then go iterate in a loop for next set of suggestions or code completions and things like that. We've got, um, an AI, uh, product, uh, called Dragon that's for physicians, and, uh, we saw a massive difference from when we used, you know, synthetic fine-tuning to when we annotated 600,000 patient, um, physician interactions by experts and actually fed that into the model and continuously optimized it to then produce, like, you know, w- I think we're sitting between 30 and 60 character acceptance rate depending on the run to something like 83%. And so that required a small group of individuals, not a large organization, that were able to actually iterate in this loop across functions and kind of all of those lines dissolving.

    6. LR

      That's super interesting. So kind of what- what I'm hearing here is if you can gather data on how things are going and then spend a l- a lot of time creating high-quality labeling to feed back into it to fine-tune it is basically the big advantage, is- is how you win in- in a lot of this stuff.

  7. 16:24 – 19:34

    The future of user interfaces: from GUI to code-native

    1. LR

      Okay. Along these lines, something else that you told me that you've been noticing that I want to hear more about is the shift from GUIs, and you kind of referenced this, from GUIs to code-native interfaces.

    2. AS

      Yeah.

    3. LR

      Talk about what that means, what that looks like, and what this means for folks building products.

    4. AS

      I think it kind of goes back to what does it mean to kind of be a product maker in the future? I think that everybody's instinct is, like, is a- is a, uh, GUI, but if you kind of think back in history, like, databases kind of went from the desktop kind of down into SQL. I think cloud was all about consoles and now it's about Terraform, and so I think we're literally just seeing the same pattern that's played out in history start to play out in AI, and like everything else in AI, it's, like, Moore's Law and it's getting faster, and so I think that's just accelerating, and if you think about, like, a stream of- of text just connects better with LLMs, and so I think that there's a bunch of trends that are kind of working in the favor for it, like the future of products being about composability and not the canvas. And I think that product makers really need to rewire their mindset around this, because I think we spend an inordinate amount of time thinking about the- the UI of something rather than how something composes, how an agent's going to be able to read something, how do you actually get infinite scale, how does that collaboration start to work? And so I think it's just a new, uh, way of thinking even though it's long been a trend, uh, that's happened in these changes.

    5. LR

      So is the prediction here that it's, uh, terminals, like, uh, Claude Code sort of experiences, or is it that it's agents that are taking it, what, or is it both? Is that kind of what you're just saying?

    6. AS

      I don't think it's, I, yeah, I do, I mean, look, if I, if that's any of us

    7. NA

      (laughs)

    8. AS

      ... knew, that would be amazing. Um, I just think that the reason why terminals and, are- are, uh, great and it feels really great when you code is because of the way it can interact with an LLM with the text stream. And I think that both can be true, that humans will continue to commit code and will find, you know, new ways to actually do that, whether it's in the IDE, whether it's in GitHub Copilot, whether it's in, you know, some new development environment, um, and I think that we'll do that with agents and agents will do that with each other and will continue to kind of evolve from there.

    9. LR

      We had, uh, Brett Taylor in the podcast, founder of Ciara, and he had a similar prediction that all software companies are gonna become agent companies.... and it's essentially what you're saying here, is that, like, your software will just be this thing that's running in the background and there's much less of a GUI. Do you think it still becomes like this chat interface, the way we're kind of getting used to? Is that like the primary interface with agents, or is there anything, something else happening there?

    10. AS

      Look, like, I think that conversation is a really powerful interface. I worked on messaging. I think it's, um, it's- it's great for lots of forms of communication, but it's not the only form of communication. I mean, we use email today to- to collaborate with each other. We use docs, like everybody uses Word and PowerPoint. Um, w- you know, there's a billion people living in places of artifacts that I think can become, uh, really important composable pieces of the picture, and I think they should be. So I- I'm excited about that. I think that, um, chat- chat will be important but, uh, certainly not

  8. 19:34 – 22:58

    The rise of the agentic society

    1. AS

      sufficient.

    2. LR

      What's interesting is ChatGPT, the number one fastest growing product of all time, maybe the most important consequential product of all time, is chat. (laughs)

    3. AS

      Yeah, it's great.

    4. LR

      It works.

    5. AS

      Um, I think the question we have to ask ourselves is, will it only always be chat?

    6. LR

      Yeah. Yeah. The way Nick described it is, uh, we're in the MS-DOS era of, uh, of- of ChatGPT and there's a Win... Which is interesting. It's like the reverse of what you're saying. So it's like, maybe if you start as that and then you have to move to GUI and then maybe it'll go back. But he said there's gonna be like a Windows version where it's much easier to understand what the hell's going on.

    7. AS

      Yeah. I mean, look, like, I think that it's- it's smart. You should, uh... every company should be bringing AI to where their users are, and ChatGPT has all of their users using chat and it's a phenomenal product. Um, and we've got lots of people around the world that do work in many different ways, and we should be thinking about how we use AI to enable that.

    8. LR

      So let's talk about agents. You've spent a lot of time working with agents, building agents, helping companies build agents. You have this really great quote that I (laughs) I love. You said that we're just starting to scratch the surface of what an agentic society actually looks like. Uh, I just love this idea of an agentic society. What is that? What does that actually look like in the future?

    9. AS

      Oh, gosh. I mean, it's- it's funny. Y- you were telling me about your two-year-old, and I have... my son, Rome, just turned one, and I can't even imagine life at two 'cause I'm just like, that is so far away and what will, what will have been developed. Look, like, I think that in the future, uh, work will look really different. I think that we're approaching this world in which the marginal cost of, um, a good output is approaching zero. And I think when that happens, we're going to see ex- exponential demand for productivity and output. And I think that the way that you scale to that is with agents, and it's agents that are embedded and they f-... they have tools and they're pieces of software, and I think there's going to be, uh, a ton of those, far more than s-... the software that we use today. And then I think there could be a set of embodied agents that are developed. And we start to see that now, right? You can assign a pull request to Copilot. You can, uh, create a software development rep that's agentic that can kind of do some of the lead generation and mining for you. And so, I think that when all of that happens, the work chart, uh, the org chart starts to become the work chart. I think that tasks and throughput become more important than, than they have been before. I also think that you just don't need as many layers. Like, I think the whole kind of organizational construct might start to look different in a few years. And so, uh, I- I'm pretty excited about it. I think, I think meetings will still be meetings and- and they'll be weird, uh, but I think they will be a bit better. Um, and I think there'll be lots of changes. I think that for the average employee, my hope and- and kind of my optimistic view is that they will be able to expand their skill set because now they have their own agent stack that they can bring with them to work, just like you can kind of bring your own device and you can, um, start to- to have access to a set of skills that you never had before. And so if you think about, you know, the 20 million people that maybe sit in that, that space i- across America and they get 20% more skilled, that's like pretty exponential for GDP. And so,

  9. 22:58 – 26:24

    The “work chart” vs. the “org chart”

    1. AS

      it's- it's pretty fun.

    2. LR

      This comment you made about the org chart... the work chart becomes the org chart is such a profound concept, because I don't know if this is what you meant, but what I'm imagining is you build these teams and here's your mission and goal and KPIs, and it's humans and like, "Oh, cool, go do this first," and what I'm... what I'm recognizing as you're talking is like, okay, but if you have agents doing that, that is their prompt, "Go drive conversion," and then you have all these agents and that's the org part. Your jo-... this is the w- the conversion/onboarding team, and that's like a bunch of agents just off doing their work. Is that what you mean?

    3. AS

      Yeah. I mean, yeah. I think like today we think in terms of, hey, who reports to who in the org chart and who's responsible for these areas? And I think at the end of the day, when you have a set of capable agents and people are capable of more things, you're not going to start to think in hierarchy and communicating upward. You're going to start to figure out like kind of outward task-based type of opportunities. I think that humans will always decide in organizations how AI is used and what we want to apply it to. But yeah, it's- it's kind of exciting when a- a new issue comes up or a new task comes up, how do you actually automatically decide where to route it? Who's working on that task? How do you actually go work on it? How do you observe if the agent's doing the right thing? How do you fine-tune it if they're not? Like all of those things. So I think that, um... I'm just speculating, right?

    4. LR

      Yeah.

    5. AS

      Uh, uh, but there's a world in which that could be pretty exciting, and I think that's great because we can just accomplish more.

    6. LR

      You touch on this point that reviewing the work is gonna be increasingly important. If you have like a thousand agents off doing work, it's just like, holy moly, that's a lot to look at, make sure they're doing the right thing. How do you think that evolves, just like being able to scale your ability to review the work that's being done?

    7. AS

      Yeah. I think that, um, the same kind of loop that we talked about becomes increasingly important. Like fine-tuning and self-healing, observability, really good evals, all of that. I mean, the good news is that there are systems that manage this for billions of people today that already exist and so I think that, you know, we don't have to reinvent the wheel. There's certainly going to be a bunch of new things to learn if that world ever plays out. Um, but I think, you know, uh, managing devices and policies and group access, all those things are solved problems, uh, which is good.

    8. LR

      This episode is brought to you by Fin, the number one AI agent for customer service. If your customer support tickets are piling up, then you need Fin. Fin is the highest performing AI agent on the market with a 59% average resolution rate. Fin resolves even the most complex customer queries. No other AI agent performs better. In head-to-head bake-offs with competitors, Fin wins every time. Yes, switching to a new tool can be scary, but Fin works on any help desk with no migration needed, which means you don't have to overhaul your current system or deal with delays in service for your customers. And Fin is trusted by over 5,000 customer service leaders and top AI companies like Anthropic and Synthesia. Because Fin is powered by the Fin.ai engine, which is a continuously improving system that allows you to analyze, train, test, and deploy with ease, Fin can continuously improve your results too. So if you're ready to transform your customer service and scale your support, give Fin a try for only 99 cents per resolution. Plus, Fin comes with a 90-day money back guarantee. Find out how Fin can work for your team at fin.ai/lenny. That's fin.ai/lenny.

  10. 26:24 – 28:23

    How Microsoft is using agents

    1. LR

      So a lot of this, it feels like it's in the future. I know a lot of it's already happening. People are using agents in all these different ways. Is there any way you or your team have found th- a value in working with agents of some kind, other than coding, I imagine, is a big part of it, but just anything there that's like, wow, that's a big deal.

    2. AS

      At this point, we have AI and agents in many of our workflows. Like, one of my favorite ones, so right now, our, uh, my engineering partner's out, so I jump on the, the live site bridges when something goes down and, you know... As something as simple as, like, you can automatically get a summary of everything that just happened 'cause usually there's 15 people talking, you don't actually know where the incident started, where it's going to end and everything, and then all of a sudden I have that and I can kind of figure out and ask questions and get updates, like, awesome. Like, I think that, um, kind of the- the entire kind of DevOps areas is, is changing. We use it to, uh... We use Spark to create prototypes, so everybody on the team is expected to code, but, like, you know, sometimes just chatting and, and, like, talking in real words actually gets you to a prototype that's more interesting and, like, more expressive and reflective of your creativity, so we use that. I mean, I think everybody's using AI to write. Um, everybody's using, uh, AI to kind of, uh, find ways to have efficiencies and, like, coming up with, uh, documentation and things like that. And so I think it's everywhere, which is cool. Um, I think that we're just scratching the surface, though, for kind of, like, what's possible in terms of working with agents.

    3. LR

      That's how I always feel when people ask me how I use AI. I'm just like, "It's just, like, everywhere. It's just, like, in every little... Sprinkled in everything I do now." I don't even know how to describe it. (laughs)

    4. AS

      Yeah, it's hard to remember a world where it didn't really exist.

    5. LR

      Yeah. There's, there's a product manager that I collab with, uh, Peter Yang, who talks about how he just doesn't... "I don't even know how to do a strategy doc anymore without AI." (laughs) Like, how did people do this without having someone-

    6. AS

      Do you think there will be-

    7. LR

      ... give me feedback?

    8. AS

      ... strategy docs in the future? That's gonna be interesting.

  11. 28:23 – 35:38

    Planning and strategy in the AI landscape

    1. AS

    2. LR

      I, I have this, like... I had just wrote this post once of, like, which skills of a PM job will be most replaced by AI, and strategy is the one that people are the most dep- have the biggest debate on. Like, you could argue, I don't know, if... Like, oh, let's get into it briefly. Uh-

    3. AS

      (laughs) Okay.

    4. LR

      You would think if some AI had all of the information you had about where the market's going, your metrics, your product today, it would be so good at developing a strategy for you. Many people think that's the one thing AI will be really not good at for a long time, because that's where we need all this human judgment stuff. I don't know. Do you have any thoughts?

    5. AS

      I think that some of the most consequential products in the world required a bunch of kind of deterministic, like, logical sets of inputs and, like, sparks of creativity and imagination and judgment and vision that could not be achieved without s- humans. Um, like, uh, Microsoft is, like, the vision of a software factory and creating what Microsoft did, uh, wasn't inevitable. Um, Instacart, uh, you know, there was web vans and web vans didn't work, but Instacart did work because of a different way of thinking about it, uh, that came through judgment and iteration and a bunch of things that you couldn't have learned unless you actually went through the process. Uh, you know, the iPod. Like, you, you go forward. So I think it's there. I think docs themselves, like, for every idea, for every, you know, need, will just start to kind of fade into, you know, applications and, and different artifacts in the productivity suite, which, um, you know, is just a different way of working.

    6. LR

      Yeah. Like, your, your original question, which I didn't quite answer, but I think is important, you were asking, like, do we even need strategy docs? And I guess it's just, like, somehow everyone needs to be aligned on the strategy. Maybe it's not a doc.

    7. AS

      Correct.

    8. LR

      Yeah. Could be some other artifact.

    9. AS

      I mean, if, if you, if you architect an organization the right way to keep up with AI, you know, you, you need a lear- you'll need the different alignment mechanisms than traditional ways of actually working.

    10. LR

      So let me ask you, actually, about that. So planning right now is just crazy. How do you, like... How does anyone plan a roadmap when there's just like, okay, GPT-5's out, c- get rid... Uh, what, what, what works for you for setting an actual, a roadmap and a strategy for your team? Like, how far out do you plan? How often do you have to rethink everything?

    11. AS

      I mean, I'll caveat this by saying, like, everyone's just figuring it out and it's a lot harder to figure it out when you're a larger organization than when you're, you know, much smaller and you get to kind of, uh, you know, run something yourself, and there's pros and cons to both. So here's what we do. We, uh... The company, um, historically, at least in our product teams, had kind of semesters that they planned against, so think of that as every six months there's kind of a strategy, look back, look forward, all of those things. I think that's very valuable. I think, like, the idea of six months though and really understanding what's changing out in front is, is truly challenging to kind of have a over-baked situation, and so we kind of think about it as...... you know, what season are we in? And so a season which is very uncomfortable can be denoted by a set of secular changes that are happening in the industry or that are happening from customers. And so, you know, you can think about season one might have been, like, you know, the prototyping of AI and kind of the early GPT work, and then it was all around models and reasoning models, and now it's the advent of agents. And so that can last a year, that can last six months, that can last three months. But, like, grounding everybody on the- the ethos of what are the secular changes? What are the customer problems we need to solve? What does winning look like? So everybody has that shared sense. What is the North Star metric is something that we do. The second thing that we do is that we have kind of loose quarterly OKR. So, like, okay, if we believe that, what do we need to do next quarter to actually put ourselves on a path to that? And then from there, uh, you know, teams are operating in squads and they're kind of setting out, you know, four to six week, um, goals that they're trying to go after for problem areas to go ladder up to that, you know? And especially as the platform for the company and the platform for our average customers, uh, with AI, I will say we- we go through lots of changes to that all the time. And I think we have to just have an openness that that is the business that we're in. I think the other thing is just, like, we try to leave slack in the system, not just for the unplanned, but for the- the slope. I think that we have to continuously be thinking about how we're going to disrupt the platform, um, in our thinking and what we need to be investing in to make that possible. And so we try to do a little bit of both.

    12. LR

      This is awesome. So, okay, what I'm hearing here is there's this concept of seasons and everyone's aligned. Okay, this is time for agents. This is what's happening right now.

    13. AS

      Right.

    14. LR

      We're gonna center around our strategy around agents, and then there's these loose quarterly OKRs you plan for three months roughly, and then you leave some slack in the system for things to change.

    15. AS

      Yes.

    16. LR

      Is the current season agents? How would you describe what season we're in right now?

    17. AS

      Yeah. Uh-

    18. LR

      Okay.

    19. AS

      It's agents.

    20. LR

      Okay. Do you have any-

    21. AS

      It's the rise of agents.

    22. LR

      The rise of agents. (laughs) It sounds like a Terminator movie. Do you, uh, is- do you have a sense of what the mi- next season might be? Is there any, like, "Oh, this might be coming next?"

    23. AS

      Gosh, um, I don't. Um, but I think that... Look, like, we have, you know, more than 15,000 agents th- that are, uh, deployed on our service today, um, at least at the- the Azure service. There's a bunch of other platforms in the company. And I would just say that I think that we should really focus on, um, making sure that we have all of the alignment, accountability, observability, evals to making those agents, like, great. I think that, uh, Manus' breakthrough in the space was that they can, like, do these tool calling loops and have agents kind of do longer running tasks that really no other platform was able to do. I think stuff like that is critical. Memory is critical. Like, there's still, uh, a bunch of building blocks that I think, like, are m- leaving agents incomplete in the wild that I think we have to really sweat the details on before we move on.

    24. LR

      So it's just like, uh, agents till the end of time until super intelligence and then we're just on beaches chilling.

    25. AS

      Yes. (laughs) Agents until dank memes. Um, look like, I s- yeah. Th- I, I think the cool thing is, is like something new could come in three months, something new could come in, uh, 13 months. Uh, I think, like, we kind of have this conviction on a set of building blocks that we wanna provide to enable these agents to, uh, endure and- and have high endurance. And so that's- that's what we're focusing on.

    26. LR

      When you said there's 15,000 agents, what does that mean? Is that 15,000 types of agents you can use or is it like that's how many processes are running?

    27. AS

      No, that's, uh, that's, you know, customers. 15,000... I, I think I should re-reference the numbers. 15,000 customers who have produced agents. I think the number of agents is actually, like, millions.

    28. LR

      15,000 customers that are building a specific kind of agent on your-

    29. AS

      Correct.

    30. LR

      ... platform and they're running... And the number of agents is in the millions just running there-

  12. 35:38 – 39:31

    The importance of platform fundamentals

    1. LR

      So let me just kind of go in a slightly different direction. You're kind of in this- you're kind of in the center of the storm of a lot of AI, just like seeing everything that's going on. Is there something you wish you'd known before stepping into this role that you're just like, "Okay, I see. I didn't expect this."

    2. AS

      When I first took the role, it was kind of described as, like, the belly of the beast, and I, um, had spent most of my career building products at the center of machine learning and applications or businesses. Um, and I think that to my surprise, a lot of the learnings have translated in terms of what makes a great platform, um, is what makes a great product. Uh, so... And, and, like, the thing for me is, like, it's often in the invisible work or the, like, not the pixels that actually drives that. So like, for example, o- one of the first companies that I worked at was a company called Porch Group. I was employee seven, and we knew we wanted to help people take care of their home and I think we invented so many features like the home report or, like, a way to manage your home or, like, house style inspiration where you could like see all of the- the houses and just ma- map every single room and... The single most important thing that we could- could have done and did during my time there was create a matching platform that matched the six million professionals with the 1,300 service types and the, uh, 37,000 ZIP codes and all of the homeowners in North America to actually take care of their- their home. And that was just a game of inches and kind of optimizing that engine in order to create higher quality leads essentially. That's what got us to the- the first 500 billion, uh, $500 million valuation. That's eventually what we built on to actually have other vertical services, uh, and software platforms that- that IP of the company. Same with messaging. Um-... the number one learning that I had was, look, like, WhatsApp didn't win because it had stickers or stories or dark mode. In fact, I, I don't even think it had all of those things when it won. It won on a few premises, because one was the phone book. Like, you knew that when you use WhatsApp, you could reach every single person because you had their phone number, and those are the people that you care about when you're using messaging. It was the reliability and how fast it was. Like, I could text my grandmother in India and know that she would get my text message all the time. And then it was the privacy. Like, when you are sending 200 messages a day to the four people you care about most, you wanna make sure no one else can read the messages, and so the end-to-end encryption really mattered. And so, it wasn't the hundred- hundreds of features, it was all in the, kind of the infrastructure and the platform. Same with Instacart, right? Like, there are so many loved features of Instacart, but at the end of the day, it's a billion items that updates 3,000 times every single minute to get homeowners their groceries from the store that they love. And so I think I, I wish I had known that, 'cause I think it would have curtailed my learning curve to say that it's not all the features for the, the platform that matters, it's the data residency so the hospital in Germany that's fine-tuning a model can do so in confidence and the data isn't going to leave the region. It's the availability, it's the reliability, it's, you know, making sure you have the right selection of the tools an enterprise needs- need and the right way to retrieve the knowledge. And that's kind of the, the platform that we've built but just didn't fully have that picture that those learnings would translate.

    3. LR

      Hmm. That's really interesting. So, what I'm hearing is people kind of undervalue just how the sim- the simple bottom of the Maslow hierarchy of, of things you gotta, of, of things that help you win in, in platforms especially, and messaging platforms including. So it's like reliability, um, privacy, I don't know, availability.

    4. AS

      Yeah. Performance, reliability, privacy, safety, um, all of those things.

  13. 39:31 – 42:10

    Lessons from industry giants

    1. AS

    2. LR

      Hmm. Let me ask you kind of a co- totally different question. When, uh, we were gonna record this previously, and you were like, "Oh, I have a big meeting with Satya I gotta do instead," uh, and so-

    3. AS

      (laughs)

    4. LR

      ... we moved it to a different time. Uh, very few people get to work with Satya. He's quite a, quite a successful leader. What's something you've learned from him about, I don't know, leadership or product building?

    5. AS

      I've learned that optimism is a renewable resource. Um, like this company for 50 years has had, you know, every reason not to succeed, and it has. And, uh, even as it's had early success in the AI era, and challenges, and other successes, like, and the space is developing so quickly, I think that his ability to generate energy and to use his optimism to kind of renew everybody's dedication to the mission, um, is unbelievable, and I think it's such an important part of the culture. Everybody talks about the growth mindset. That's a real huge part of the culture. But I think the ability to, to generate energy and clarity on what we need to go do, and, uh, use optimism to renew the commitment every single day for every single person in a entirely competitive talent space, is like, i- is pretty amazing.

    6. LR

      Is that something you think that is just innate to him or it's something he's worked on to just generate this optimism on behalf of everyone?

    7. AS

      I have no idea. (laughs) Um, we should ask him, but I am, like, deeply impressed by it.

    8. LR

      It's interesting that a lot of this comes down to just vibes.

    9. AS

      (laughs)

    10. LR

      It's just like this vibe of, you know, like imagine it's not him just the words he uses, it's just like this energy that he exudes, optimism and energy.

    11. AS

      I mean, think about it, we all choose to, to, you know... Someone here just said this to me and I thought it was great, we all choose to close the door on our kids every single day to go work on something.

    12. LR

      Mm-hmm.

    13. AS

      And so you have to work on something that is, like, deeply moving to you and is, like, you know, you have a deep belief that is going to make the world a better place and, like, I think that's why it's vibes. Like, I, I think you, you have to follow, um, a- and have a sense of duty towards a mission that is bigger than yourself.

    14. LR

      That makes me think of a line that I've referenced a couple times on this podcast that's really hits people really hard that, "The only people that'll remember you working late are your kids."

    15. AS

      Okay, I don't know where we're going with that (laughs) , but that was like, you know, now you're like-

    16. LR

      It's too much. It's too much.

    17. AS

      Yeah.

    18. LR

      We've gone too far.

    19. AS

      Gosh. Ah.

    20. LR

      (laughs) Oh man. Okay. Well, let me ask you this. What's, what's driving you-

    21. AS

      We could've, we could've said our customers, we could've, we could've gone a different route on that one.

    22. LR

      (laughs) This is the real, the

  14. 42:10 – 44:30

    What’s driving Asha

    1. LR

      real stuff. Um, what's driving you? What's driving you? What's keeping you excited about, uh, the work that you're doing?

    2. AS

      What AI will help us do from a workforce perspective, what it will help us do from a healthcare perspective. Like, you know, my mom has cancer and I think a lot about how, wow, we might find a way to solve the form of cancer she has in my lifetime, and I never thought that was possible three years ago. Like, all of that's deeply profound. And the thing that, like, I personally think a lot about now that we know that we're living in this time working with such powerful technology is the, the effects of it and how I can, you know, best build a platform where people can make use of it. So, like, the reason why I work at Microsoft is because, like, the whole ethos of the company is like, how do I help people and businesses achieve more? And like, more for me and the thing, like, I think about at night, um, outside of, um, you know, GPUs, is, um, you know, I, I think about, like, will my son have classmates in the future? And that's not because agents are going to replace them. It's because the, the fertility rates are declining, right? Like, the, uh, the average birth rate in the '90s when we were growing up was like three, and now it's 2.3, and in 2050 it's estimated to be, you know, below replacement.And I think that AI can have such a big effect on it, and already is. Like, I was just reading about a hospital in London that's, you know, able to improve pregnancy rates by using AI to match, you know, eggs and sperms, and they're cutting cost at the same time. Uh, y- you saw with the ChatGPT-5 launch yesterday, such an amazing story about how ChatGPT is helping in healthcare. You know, Stanford's one of our big customers of the ... with the platform that I, uh, build, and they're working on using AI for tumor reviews. And, um, it, it's just, like, th- that is, like, it is these sets of things that will, like, move humanity forward and expand our lifetime, and give us the, like, privilege to solve 100-year problems. And so, that's, that's why I, I'm excited, and that's why I do what I do.

    3. LR

      Yeah, especially in your role where you are building the platform that enables all of this. I could see how, uh, impactful that could

  15. 44:30 – 49:19

    Reinforcement learning (RL) and optimization loops

    1. LR

      be. Asha, is there anything else that you wanted to touch on or share or double down on, uh, of anything we've talked about before we get to our very exciting lightning round?

    2. AS

      We touched on it a little bit, but I think that, um, with the advent of agents and products that think and can act and reason, there's going to be this kind of new wave around RL. And, uh, I have a deep belief that that, that will become one of the most important product techniques, kind of, of the next season or at least the next few seasons.

    3. LR

      And RL is reinforcement learning.

    4. AS

      Yes. Yes, exactly. Like, I believe we will see, you know, just as much money spent on post-training, uh, as we will on pre-training, and in the future, more on post-training. Um, we talked a little bit about Nathan Lambert's study where his review was that, you know, when a model hits 30 billion parameters, it makes more sense to kind of fine-tune and optimize that. You know, 50% of developers, uh, according to surveys, are now fine-tuning. Uh, and we know fine-tuning is good, but like, if you actually go through the full loop, you can get better results. So, I think there's, there's a bunch there, and I think there's a whole new set of infrastructure and platforms and, uh, companies that will be created that are all around this part of the stack. And so, um, I think it's an exciting time to be in the platform space, but it's also an exciting time to be starting companies and be thinking about those problems.

    5. LR

      I wanna make sure people truly understand what you're saying here, 'cause not everyone truly understands post-training, pre-training. What's the simplest way to understand the difference there, and just why it's such a big deal that, that investment is moving to post-training?

    6. AS

      The way that I think about it is, you know, to create a foundation model, um, creates a trem- uh, i- it requires a tremendous amount of compute, a tremendous amount of science, uh, expertise, as we're seeing, who's, uh, which the cost for scientists, their (laughs) average value is rising dramatically. And I think, you know, m- uh, an expertise that, eh, we've seen, it, like, isn't everywhere in the world right now. And so, it's just a big CapEx investment to do that. And with this explosion of models that we talked about in the beginning, there's a lot of good models to choose from for different domains. And so, I think that you just get more leverage economically, you get more leverage from a taste perspective of how you actually want to steer a model, uh, if you're actually doing reinforcement learning or some sort of fine-tuning to actually start to optimize what's off the shelf for some outcome, like price, performance, quality. And if you think about that, that's, that's not crazy, right? Like, you know, ranking is an age-old, uh, optimization problem where you don't wanna just take what's off the shelf, because there's, like, amazing frameworks and UI and kind of components that, you know, the world ... there's React components that are out there. You still want to tailor the experience to a set of use cases or a set of people. I think it's just the same kind of industrial logic.

    7. LR

      So, in practice, what you're, what this means is there's, like, a GPT-5 model. You're saying there's a lot of opportunity and a much more efficient way to spend money, which is take something like that and then train it on additional custom data that you have, whether it's data or just reinforcement learning, maybe even with, with humans to align it with what you want it to achieve.

    8. AS

      Yep, and it could be your own data, it could be data that you buy, it could be synthetic data, it could be, you know, something else. But I think that, um, I, I think that we're kind of going to start to see, you know, more and more companies and organizations kinda start to think about, "How do I adapt a model?" rather than, "How do I take something, uh, off the shelf as is or invest a bunch of money in building my own models?"

    9. LR

      Yeah, I forget ... I know Cursor, w- when he was on the podcast, he shared that they have a bunch of models that, that support your experience with Cursor, and over time, they're just gonna have their own thing. I, I forget who was it, Windsurf or one of those guys just uses their own model now. They don't just plug into Claude.

    10. AS

      Hmm. I'm much more in the model system camp. Like, I believe in, uh, model diversity. I think that in experience, like, Claude, uh, like Sonnet 4 is awesome for a set of use cases, versus GPT-5 is different for different use cases. I think that there are some tasks where you care about the latency of the model, you want ... m- you're, like, cool with the thinking time, or you kinda want, uh, quick retrieval and things like that. Like, uh, I, I think the, the beauty is there's a lot of models that can kinda help you achieve that. And so, I'm much more in the, like, model system, uh, rather than one model to rule them all.

    11. LR

      Is that the right term? I've also heard ensemble model, ensemble of models.

    12. AS

      I think about an ensemble of models as a set of multiple models that then you can, you know-

    13. LR

      Mm-hmm.

    14. AS

      ... fine-tune and deploy independently, but, you know, uh, at this point, we're all making up different terminology to define things that we, like, have deep beliefs on that have, like, you know, uh, s- limited sets of data points because everything is moving so fast.

  16. 49:19 – 57:10

    Lightning round and final thoughts

    1. LR

      Yeah. With that, we've reached our very exciting lightning round. I've got five, 10 f- (laughs)

    2. AS

      I'm ex- I'm very excited for our lightning round and I'm, like, turning down the lights.

    3. LR

      (laughs) And then they'll come back on, I imagine, in one second. Okay, first question, what are two or three books you find yourself recommending most to other people?

    4. AS

      At work, it's probably Thinking Machine. Uh, so it's all about treating the, the cause, not the symptoms. The, like, you know, prototypical example is like, you know, the ...... if, if you want to solve traffic, you don't actually put up, you know, speed bumps or, or, or speed limits. You actually have to, like, solve for walkability and mobility and kind of, like, why people actually use cars. Um, outside of that, uh, I'm kind of personally, um, the CMO of Instacart, uh, recommended to me Tomorrow and Tomorrow and Tomorrow, and I read it, like, last month and last year and the year before because I love it so much.

    5. LR

      Wow.

    6. AS

      It's, like, this, like, beautiful story over 10 years.

    7. LR

      Hmm. What are some favorite recent movie or TV shows you really enjoyed?

    8. AS

      Formula 1, saw it twice. Uh, For All Mankind (laughs) . Uh, For All Mankind, I like season four. Uh, I don't know. I li- I like kind of playing out alternative theories to kind of how the space race might have looked.

    9. LR

      Do you have a favorite product you recently discovered that you really love? Could be tech, could be gadgets, could be clothing.

    10. AS

      So I just joined the, the board of, of The Home Depot and we're doing a little renovation project. And so, uh, there's this new, kind of, well, new to me, DeWalt kind of power pack and they use pouch cells, and so it's, like, 50%, like, lighter but with all the, the power, and it's, like, awesome for drills and, like, things that, you know, I need to lift up with one hand that feel heavy. So I love that. We also are testing out this new brilliance smart home kind of system, uh, so it's, like, kind of four inches of, like, high-res middleware that allows you to kind of connect to everything, and I've, like, reached peak kind of DISSAT with, like, the explosion of all the technology required to actually use your home. So it just might be the middleware that, like, sticks, but we'll see.

    11. LR

      Did you say DISSAT? Is that short for dissatisfaction? (laughs)

    12. AS

      Yes.

    13. LR

      I've never... (laughs)

    14. AS

      Sorry, I'm speaking in acronyms.

    15. LR

      Whoa. I have never heard that. DISSAT. It's like... (laughs)

    16. AS

      (laughs) .

    17. LR

      I love that. By the way, I love that you're on the board of The Home Depot. What a, what a different part of the spectrum of, of, of work, um, yeah.

    18. AS

      Uh, I, it's been, it's been awesome. The very first board meeting, the head of philanthropy been at the com- uh, has been at the company for decades and she said, "Welcome to the greatest company on the planet."

    19. LR

      Hmm. Wow.

    20. AS

      It was pretty special.

    21. LR

      You're like, "Now Microsoft!" Uh, is there something you've learned from working with that, with them that you've brought to Microsoft?

    22. AS

      Look, like, it's, it's, it's new, it's this year, but I've long worked at, on products that kind of had that impact. So, like, when I was at Porch, it was pros. At Instacart, we had 600,000 shoppers, um, and obviously The Home Depot, Depot has associates. One of my favorite things about the company culturally is they have this inverted pyramid where instead of having, like, executives at the top, the associates are at the top, and the stores themselves are, uh, headquarters, uh, of, uh, and, and then the, the kind of traditional HQ is kind of support. Um, and so it's just, like, it's so customer-centric, and when I think about amazing execution and creating these durable, long-term institutions and kind of h- how culture and ideology and, and kind of leadership is formed, like, I think about that, and I think about at the end of the day, you know, AI is going to have an impact on every single person and every single job, and it's, like, amazing to kind of just spend time with people outside of our bubble, uh, and, and kind of really try and learn what their real pain and problems and how they think about AI and how they think about technology and kind of what we need to do.

    23. LR

      Okay. Two more questions. Do you have a favorite life motto that you find yourself coming back to share with friends or your family?

    24. AS

      I, I, I used to use the kind of minimize regret framework, um, and I, it's, it's great and I've, I've used that for a long time. I think that probably once I got into my adult years and started to kind of have a family and things like that, my, uh, kind of just worldview changed a little bit and I wa- it was all about maximizing kind of option value, and it just gave the things that I naturally cared about, like family and health and trust and relationships, like, it, it was just kind of, like, a new level of, like, value associated with those, because all of a sudden, learning rest on the weekend can, like, compound in the future, or, you know, having good health can compound in the future. You don't, uh, have to trade that off of working extra hours or, you know, the importance of family and all of those things. And so, I think that, like, the, my worldview is, like, when I'm 70, it's not about what do I look back on in my life and count the number of regrets. It's really about, like, looking forward and the number of adventures I will still have because I have, like, uh, accumulated this wealth of, of skills and trust and, you know, people and, and family and impact and things like that.

    25. LR

      Speaking of skills, uh, the internet tells me that you, you're a second-degree black-

    26. AS

      Oh, gosh.

    27. LR

      ... belt in TaeKwonDo. (laughs) Uh, why, oh gosh? Is this true? And then I have a question about it.

    28. AS

      Uh, this is true.

    29. LR

      Okay. (laughs) That's incredible. What's some, w- wh- why are you, why is this embarrassing? That's an incredible thing. Uh, okay.

    30. AS

      I'm, I'm generally embarrassed anytime anything is discussed about me. (laughs) Yeah.

Episode duration: 57:10

Install uListen for AI-powered chat & search across the full episode — Get Full Transcript

Transcript of episode J9UWaltU-7Q

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome