Back to First Principles | Productiv CTO Ashish Aggarwal

0:00founders are being forced and this is good actually that founders are being forced to think like business people. They're like, look, you are creating a company, but after all, you are creating a business. You have to solve a real problem. You have to have real customers. You have to have a real plan of how you will make money. And it could be that the money will come in the long term. That is an okay plan, right? It doesn't need to be profitable from day one, but you need to think about that instead of the good old plan of, well, we'll just get more money from the market. Tell us if this problem sounds familiar to you.

0:40You want to use engineering metrics to improve your work, but every time you try, you realize there's a gap between the metrics world and your team's reality. And even with all the frameworks that already exist, is there enough context to make an informed decision about what metrics to focus on? That's why Linear B is teaming up with Refactoring. Together, we're creating a comprehensive industry report about how engineering teams use data and metrics to improve their practices. But we need your help. This week, we're launching a community survey to collect real-world stories that will help us build a practical playbook for tech leaders everywhere.

1:16The best stories will be quoted and published with the final survey results. To share your story and get a chance to help engineering leaders everywhere, visit refactoring.fm or head to the link in the show notes. Welcome back to Dev Unrupted, everyone. I'm your host, Conor Bronson. Today, I'm joined by Ashish Agarwal, founder and CTO of Productive. Ashish is not only a successful founder, but also an active investor in over 30 companies. He's someone who's deeply involved in the startup ecosystem. And today, we're excited to hear his insights on scaling teams, the impact of AI on startups, and how to navigate the current macroeconomic climate.

1:52Welcome to the show, Ashish. Thank you, Conor. Glad to be here. Yeah, I'm really excited for this conversation because I think you bring a lot of insight both from your personal experiences and the kind of awareness you have of what's happening within tech startups, within software engineering broadly. But before we jump into the conversation, I do want to remind our audience that if you enjoy our episodes, please just take a brief moment to rate and review Dev Interrupted on your podcasting app, Spotify, Apple Podcasts. Your feedback helps us reach more listeners and continue to bring you more great conversations with engineering leaders like Ashish.

2:27Now we've got that little bit out of the way. We all know the startup ecosystem has changed. It's becoming increasingly difficult for companies to receive enough funding, let alone become unicorns. In fact, there's data from Carta that shows that of startups that raised their seed round in Q3 2020, only about 40 % have graduated to Series A in two years. But that's actually quite a high rate compared to the folks who raised their seed cohort in Q1 2022. That's just 15.4 % of those folks have actually raised to Series A in the eight quarters set. So we're seeing this cash crunch. And then on top of that, Gen.ai is reshaping the tech landscape and leaders have to reckon with how to keep their teams motivated and happy in these challenging times.

3:09So with that in mind, Ashish, as someone who's deeply involved in the startup world, what are you seeing in the ecosystem right now? And how do you think startups should think about growing in this environment? Right, Karsov, you are right, right? Things have changed. I would not say things have changed for better or for worse. It's just that it's a new environment. And I think you've heard of this, that it's no longer growth at any cost. It's now the name of the game is efficient growth, right? And so that's what I am seeing in productive, but also across the market in different companies that I am involved with.

3:49Investors, and rightfully so, are saying, hey, look, money is not cheap. we want you to judiciously use the money, invest, build products which are useful, don't just go change some pie in the sky kind of thing. So efficient growth is the name of the game. Definitely. There's been a shift from this aggressive growth, growth at all costs, as you put it, to more sustainable, more efficient strategies. And it's definitely a trend I'm noticing in conversations with industry leaders. What challenges or opportunities do you see this creating for startups today? To some extent, it solves that problem of saying this, the market, the startup market, startup ecosystem had become too crowded, right?

4:33Everybody running after, you know, any reasonable opportunity. I think that's what we are seeing now. The opportunity for startups is clear, especially for founders. I have a clear vision of what customer problem you're solving. how you can actually solve it without requiring a giant boatload of money to make a small amount of money. Founders are being forced, and this is good actually, that founders are being forced to think like business people. They're like, look, you are creating a company, but after all, you are creating a business. You have to solve a real problem. You have to have real customers.

5:13You have to have a real plan of how you will make money. And it could be that the money will come in the long term. That is an okay plan, right? It doesn't need to be profitable from day one. But you need to think about that instead of the good old plan of, well, we'll just get more money from the market. We don't need to actually make money. And I think that's good for the startup ecosystem. Overall, it makes everybody more healthy. if we think of first principles in building a business. I agree with you. I think it's a positive for the ecosystem in the long run, though I know it can feel hard for founders and employees in the moment.

5:53We're not trying to downplay how hard this is for many companies, for many people, but the long-term impacts on having better businesses that are more efficient, that hopefully have a better impact for the dollars spent on them should be positive. No, absolutely. And I just want to echo that. This is not to downplay at all how hard employees, founders, even investors are feeling with this market. Absolutely, everybody is feeling the heat. But the more people I talk to, the more they say, hey, Ashish, you know what? We are that company. We are those people who are solving a real problem, who have real customers.

6:32In fact, the fact that we are forced to or we are more encouraged now to talk to the customers, to understand the problem a little bit deeper, to understand the revenue potential, to show the value to our customers and then to the markets. It's helping us build a good product and a good sales motion much faster than if we were not under this pressure. So the pressure in some sense is bringing out the best in a lot of people. Do you think that a lot of this kind of rationale of, hey, we have to be like the full winners. We have to monopolize a market. We have to just push for growth, growth, growth.

7:08I mean, I know it was related to cheap money, obviously. Like that's a huge part of it, low interest rates. But do you think part of it also was related to seeing these outsized winners and getting obsessed with the Googles, the metas and the kind of massive outcomes they had versus realizing, hey, I can run an efficient business and maybe I'm not going to own an entire massive market? It's somewhat true. I, you know, again, I cannot speak to who was thinking like what, but, but I think in general, that mentality was, I did see that kind of saying, hey, we'll figure out the money making part later.

7:42People were always, I think, I think they were disciplined enough to say, hey, are we solving a good problem, a real problem, or at least a realistic problem. But that part about efficiency wasn't part of the picture because nobody was asking for it. So like, go keep spending money, get to, you know, get to 100 million customers or something and then figure out how to make money, right? And only a very few people in the world or a very few companies in the world now can afford that. I mean, Facebook can clearly still afford that. It doesn't make any money for WhatsApp, but hey, it's a good business.

8:15Brand play, yeah. And play. But not everybody is Mark Zuckerberg, so, you know. Very few, yeah. So we have to have a different strategy than Mark. So given that need to adapt to more efficient growth, are you seeing more teams looking for talent from non-traditional backgrounds or in non-traditional geographies? Yes. And to tell the truth, I am hearing that this is a trend not just in small companies or startups, but also a trend in much bigger companies. I'm hearing that most of the new budget that comes in is first allocated to, quote unquote, offshore growth, whether it is Latin America or India or wherever, right?

9:03Wherever is your favorite location for getting the talent. And I think a lot of factors are combining to that, right? One, we all know that, you know, COVID as bad as it was. it also showed all of us how we can work remotely from anywhere right yeah and so that opened the to say well you know if you have good talent in fact the top talent especially knows how to work from anywhere in the world the tools have become more mature the mentality has become more mature of working with each other from anywhere so the teams can be geo-distributed and so now Now, and because of the cost pressure, now the name of the game has become, you know, get the top talent.

9:50Because after all, that is also part of the pressure to say you have to solve a real problem. You have to solve it fast. You have to make sure the customer value is there and so on and so forth, which means you need top talent. But the good news is there's talent all over the world, including in the United States, by the way. We are part of the world and we have a lot of top talent right here in the United States. but it doesn't need to be limited to the locality that you happen to be in. So yes, I am seeing startups and big companies, but especially startups move more towards geodistributed teams instead of central co-located teams.

10:29And to your point, that's really been enabled by us being forced to figure out working remotely. I had been working remotely or hybrid for years before COVID, but now everyone has had to make that transition and at least experience it for a couple of years. And it's really, I think, made companies aware of the opportunities to acquire talent worldwide. And it's made the job market competitive worldwide, it seems like, as well. Absolutely. Absolutely. Yes, this is happening. But also I'm seeing on the opposite side, there is still benefit in pockets, right? Depending upon the role, depending upon what you want to do, what problem you are solving specifically.

11:06Some roles, some teams just work better when they are co-located, when they are brainstorming on a whiteboard. It just makes higher efficiency or higher productivity for the company. Again, reminder that the problem we are trying to solve is just higher efficiency for the company, not geodistribution for the sake of geodistribution. It is really the top talent which is driving things, not just that you can get things for lower cost. Nobody wants to sacrifice productivity here. So in some cases, I'm also seeing that people are saying, it can vary from company to company, situation to situation.

11:41But these kind of jobs should be co-located because they just work better together. I'm curious if you think that is true of like product and engineering roles, because there are many C-suite leaders who are kind of emphasizing return to office. So I'm curious on your perspective there. Yes, product and engineering is interesting. I mean, it's unfortunately not a single kind of a role, right? Product engineering has dozens of sub-rules in it. So for example, when your web engineer or front-end engineer is trying to work with your UX designer and they want to brainstorm things and if they want to use a whiteboard, which they typically do, and co-create, does it help if they are in the same room compared to a Zoom call?

12:28I think so. I think it helps that they are in the same room. right but if your infrastructure engineer who's trying to you know deploy the latest version of the os on your aws machines right and in fact even even their teams the infrastructure team do they say that hey man will work well even if we are remote in fact we might work better if you know if we are uh if we are free of distractions and nobody is like interrupting us so it it varies even within product and engineering, what kind of role? And by the way, even those roles in some other company might be flipped. Somebody might say, my infra team needs to be in person and my product design can work remotely with each other.

13:13I empathize with the leaders who say, back to work is necessary for some roles in some situations. Hell, I am one of those leaders who says, not everybody needs to be co-located. We have a geo-distributed team here at Productive. You know, we have engineers and other people in India. We have people in New York, in Denver, in Seattle, in Palo Alto, and so on and so forth. So we are a geo-distributed team. But then within these hubs, when I see some people, you know, come together to work and brainstorm ideas, you know, whiteboard or not, I think they often call me back and say, you know, it was a good idea to meet at work, you know, a couple of times a week.

13:58a me to school faster yeah it's really interesting as someone who i think personally i thrive in a remote work environment but i like even as someone who is very pro remote work in general there are clear benefits to in-person collaboration there are clear benefits to co-location at times and like i've been lucky to work in an environment the last several years where i am generally remote but then i'll spend a week or two every quarter in person with my team in person with my colleagues, collaborating, collaborating across teams, collaborating on like key projects. And then we go out and execute on that work afterwards.

14:36But to your point, I think it does really depend on the culture of the company, the type of folks you have involved and the type of roles. And like, it's a really tough balance right now to both try to acquire global talent, get access to all these incredibly talented people worldwide, while also retain some of those collaborative benefits and find the right mix. so yeah I think your your point about the variation between roles is one that I think a lot of folks need to hear in what too often becomes an almost like religious debate of remote versus hybrid versus in office and there's one or the other instead of this like series of grays which I think is most things yeah it's true and you know for part of engineering product and engineering my philosophy has evolved into to reuse the word a hybrid philosophy which is It's not pure remote that, you know, if there are whatever, 100 people in the team, they are in 100 different locations and, you know, everybody's just pure remote.

15:36But it's also not that all 100 are in the same location. So what's the hybrid? The hybrid is create, you can call it hubs, create pockets, where you put a team together of the right roles and say you people can collaborate in person, again, in multiple times a week or whatever works for you. And you have this certain goals that you are trying to solve. You have a certain problem that you are trying to solve. If you distribute this properly, you'll create multiple hubs and each hub is solving an important problem and they have the choice to collaborate in person if and when they need to. And I think that works a little bit better, more flexibility on both sides.

16:18You get the global talent, but you also have the choice of working in person, especially the product and engineering. I think that works pretty well. Obviously, there's this massive alchemy people are trying to solve right now of how do you set up your teams for this new geo-distributed world? with global talent? How do you make sure you retain some of that in-office collaboration? How do you set up the right infrastructure? And so it's awesome to hear about your approach to this. But there's this entire other element of change that's coming through too, which is obviously AI. You know, speaking of efficiently scaling a business, I'd be kind of remiss if I didn't bring this up.

16:51But it's interesting because if you ask, you know, individual developers, many will say, at least senior devs, I don't really use AI that much. Like it's not, it's not helping me with the things that are actual problems for me, like meetings. It's not helping with the things that make me less efficient, which is some of the, it's not the coding that's the problem for me. Now, I think you talk to junior devs who are coming up and they're like, oh, I rely on this. It's very important to me. And if you talk to leaders, I think leaders are all think there are massive opportunities with AI. And even many devs will talk about, you know, tests are a great thing that I can, you know, I can offload out to an AI to help me write some tests and do these other things.

17:24What's your perspective on this kind of AI wave? How are you utilizing AI productive? I see so many questions in one, right? So I will, I will, I'll take them on its time. I think is AI useful. Absolutely. Right. Is there a disconnect between, you know, how individual developers might be using AI to, to improve coding or testing or what have you versus what's the, versus what's the marketing pitch saying, hey, I will solve world hunger. Absolutely. So I'm, let me agree with you. The promise of AI today is way bigger than the actual reality of how people are using AI, including developers. Has AI made productive developers 30 % better?

18:12No, it hasn't. Have we tried looking at it? Yes. In its current form, we said, let us wait a little, right? We are not finding it as useful as it's hyped to be. I'm hearing the same thing from many other engineering leaders saying, as a leadership level, we want to try it just in case there is something here. But individual engineers are saying that trying is one thing, but I have my day job here. It's not helping if I have to review the entire output of Gen AI to see what did it miss and what did I get. And plus, I have to maintain all the code anyway. So I have to understand the whole code and have to correct mistakes.

18:54So it helps, but not really that much. And we have explored the ideas behind testing, test generation, or code reviews, or code generation, actually. Is it going to get there? I think so. I think AI is one of those things that we are an armistice, right? And as we are building and learning, you were asking earlier, how is AI impacting the startup ecosystem? It's a sword or it's a knife that's cutting both ways. On one hand, every investor says, if you are not a pure AI company, well, then, you know, I need to think twice about funding. Not everybody's thinking I like that, but many are. Plenty are.

19:35We could name some names. Yeah. Yes. But on the other hand, AI is such a giant opportunity. Right. Some people were saying to me, you know, some founders were saying to me, like, this is the wave of Internet all over again. And sure, it's early days. We don't know how internet will play out. You know, if you remember back in the 90s, we didn't know how it would play out. But eventually, we know how it played out. So is it this wave of internet that will completely transform the world? We might be in a very early stage of that and people will figure it out. Or it could be one of those waves that did not play out, like a blockchain wave where we thought it would change the world and it didn't.

20:15I personally think AI is one of the internet kind of waves, right? I think it will take a few years for us to figure it out, but we will figure it out and it will enable all of humanity, including engineers, including developers, to do their jobs in a different way, in a better way. So Gen.E.I. especially, apart from on top of AI, I think is a really, really important innovation that has happened. And so it will help us. How are we using it at Productive? I think that's one of the questions you asked. Yes, there is a general purpose thing that everybody knows that, you know, you use it to write marketing emails or you use it on the go-to-market function side, whether it's customer support or whatnot.

20:57So we are dabbling with that, some of those tools in there. on the product side, we are more involved with AI and Gen AI in building our product, or rather in the product features we enable for our customers instead of having our developers use Gen AI to write code. So we don't do that. The second part, our developers are not using Gen AI to write code. Everything is done literally by hand in the old fashion. We are using Gen AI to provide human-friendly or human-friendly interfaces to interact with our product so that people can, our customers can benefit from Gen.AI to understand our product better, to get insights faster and better, to get their jobs done faster using product.

21:46Yeah, I wonder if right now we're seeing the most efficiency gains when you have enterprise scale with AI on the product side of things. Because I look at like the recent announcement by Amazon that they see their generative AI tool internally saving, I believe it was 4 ,500 years of work and$260 million annually with it focused specifically on code transformation capabilities around taking them from like Java 8 and 11 and applications to Java 17. These kind of things where it's like, okay, this is a migration. Devs really don't want to spend their time on this. Most devs are not excited by that.

22:21But it's something that we want to get done for efficiency's sake long-term. My take right now is that, to your point, a lot of these tools that we're leveraging right now in AI are so experimental, especially on the code side, that in order to have the challenge of making sure it all works and the trust side of it, the safety side of it, makes sense, you need this level of scale where you can have a pull plow from engineering team that's devoted to that and say, hey, look, yes, we're gonna take this first pass with AI and then we're gonna review it. And it's gonna save us a ton of time if you have the scale to apply that.

22:52But for a team of 100 engineers, it's a lot harder to have a bunch of folks you can throw in a problem like that when you still have to deliver other business critical work. I think you're on to something there. When you have a large amount of repetitive work, which you can say there's a standard way to do this. We can teach an AI agent on how to do this and then do the repetitive works. It's better to do it via AI than just plain old automation that we used to do in these cases. So yes, the power of automation or standard repetitive work has been, I guess there's more power if you use AI to do that work.

23:28But where most of the work is net new, whether it's uncharted waters with a typical small development team, you have that. Then it's harder to use AI as of today. It might get easier as we go along, but harder as of today. Yeah, to your point, I think the innovative work, the kind of innovation piece of technology and startups is where AI is not yet as strong, from my viewpoint, at least. I agree with your point. You did say you think AI is a transformative internet level wave, though, potentially. What do you kind of see as the future of AI as we go into the next couple of years, as maybe it gets better at solving those technical problems, or as it gets applied in other areas of technical expertise?

24:12Absolutely. So there are multiple layers of Gen AI and AI. And we have all seen those charts, I think. You know, the core of LLMs, which only very few big companies can build. And then there are two on top of that. But then the outermost layer in some sense is this, you know, creating agents. And the simplest way to say this is take any human or any human doing any job and say, we will have an AI agent who can do this job. We can have an AI agent who can interview the interviewers. or we will have an AI agent who will respond to you, Connor, or what have you, right? I mean, everything. And I think that is, now, we may not be able to replace humans completely, but I think what will happen is as AI gets better, we will be able to assist almost all humans in almost all jobs that they are doing.

25:04You know, not literally all humans with literally all jobs, But most humans in most jobs will say, we have an AI assistant and it makes us go faster, whether it's in generating content, whether it's in generating insights, whether it's in understanding the world, whether it's in operating machinery or actually doing things via robotic arms or what have you. Yeah, I have to say, I think Microsoft's kind of perspective here about like not just co-pilot for engineering, but co-pilot for every kind of role is one that makes sense to me broadly. So I'm really interested to see how that is continued to be implemented because I agree.

25:41I think there's an opportunity to create those efficiency gains for folks over time. It's just a question of, you know, when do you get the scale to make it work? And when does it get strong enough to really be impactful? And that obviously will vary depending on job areas, depending on how things are set up. and i hope that it will kind of reduce the need for some technical expertise in using complex products at times giving letting folks learn faster and i'll say like just on a personal level like leveraging ai to help me learn something faster has been a great use case and so i can see how that can potentially scale across a lot of different areas no absolutely you know when When humans are freed from, let me say, the mundane task or the repetitive task, or even some of the creative tasks, but if somebody else is helping them, in this case, a co-pilot, my belief is that humanity just ups, levels up and says, now that I have free time and my mind is free to do more innovative things, we just do it, whether it's learn faster, whether it's do the next level of things.

26:46So it will be good for innovation all along that some tasks are punted to AI co-pilot to do for us. And learning is absolutely one of those things where it becomes a lot low pressure, in my opinion, when you are talking to an AI-based teacher instead of somebody who's pounding you over the head as one example of learning. yeah i'll be fascinated to see how it transforms the educational sector uh over the coming years because there's a huge opportunity there to improve a lot of learning outcomes if done well absolutely i and again none of us are saying that it will replace teachers right uh not saying that because there's a lot of human element of you know an emotional element of how a teacher grows a child or whoever is trying to learn.