Meraki's mission is to help connect the next billion people to the internet. These customers took the product to the far reaches of the world. That was the power of the internet. As engineers, we love building things. We love fixing things. But really, we love doing things that are useful. And so, this was the spark that led us to Samsara. Energy, utilities, logistics. It's all part of the shared infrastructure of our planet. We realized that we could bring our technology background to this world of operations and infrastructure and be useful.
Welcome to the stage Sanjit Biswas.
Maybe where I want to start is you've been part of some of the best infrastructure companies that got built out in the last two decades, but you started off first in research and then actually made it into a real product. How does that happen? Because so few people can make that transition from yes, this— the video can answer the question. [laughter] But so few people can actually make that transition. How did you guys do that? Going from deep research in your PhD to building an amazing company.
Yeah. So first of all, thanks for having me. This is an awesome event. It's fun to be around networking people and so much fun new networking gear. So my background is I studied computer networking as an undergrad and then a PhD student at MIT. We built a really big network which was called the MIT Roofnet project. This is like 20, 25 years ago. And for us the transition from academia to industry was actually kind of happenstance. We just wanted to do a cool project. And the project was take the ideas behind Roofnet and make it easy for anyone to build a big network. And the best way we thought to do that was to literally put it in a box and ship it out so other people could do it. And our original startup idea wasn't to do a startup. It was just to get the stuff out there. So it was a little bit of an unexpected transition, but then once we started getting the stuff out there running the customer feedback loop, it just took off and it was a lot of fun.
I think the other thing if you talk to a bunch of the folks that have worked in teams that you've built is what you just mentioned on that customer feedback loop. So you do that in hardware because it's you know it's not malleable.
Yeah. It's brittle and changes take a while but you've always seem to have figured out even at Samsara today what's that loop like and how does that work?
Yeah so you know hardware is quite different than software in terms of like it's long-lived, the design cycles are long and there's lots of other characteristics like thermals and things like that that aren't obvious and intuitive to a lot of folks. And I think you guys are living this too which is you want to be able to iterate. But at the same time you can't just like turn software features and do feature flags on hardware and stuff like that. So what we did is we basically try to gather a lot of data. We would always go spend a lot of time with customers. We still do that at Samsara today and we looked for the 80% of the 80/20 right what is a common feedback that you're hearing and you have to have really sort of big ears and big eyes and look around for like okay what am I missing here? What's the common thread? And then what's interesting is when you're looking for the 80 and the 80/20 it's pretty obvious right? You're like, 'Okay, it needs to be able to meet this spec or it needs to be deployed in this other environment or I didn't realize that industrial environments have all this water or whatever it is.' You start to see the common pattern. So the key is to gather the data. You can't just do it in a conference room. You have to get out in the field.
You always seem to be going after legacy and legacy industries.
What is it that about that it attracts you and how do you see that change happening over these industries? Because even what Samsara has been tackling is pulling a legacy industry into like the modern times, but how do you think about doing that?
Well, you know, I don't think it's necessarily about legacy industries as much as it is about how do you maximize them? And some of these legacy industries, if you will, they're massive, right? And so at Samsara, we serve the world of physical operations. These are utility companies and construction companies like just large scale operations and many of them have been around 50 or 100 years. So they are sort of legacy industries in that sense and in certain regards they're set in their ways but they're also very open-minded in that they have problems they want to see solved right and I think when we were in networking it was kind of similar where networking had been around for a while right it had been around 25, 30 years at the point that we entered it but there were some challenges there especially in terms of keeping up with demand right and that's always the case in technology is there's a new set of challenges and that's what I find exciting to solve is like what are the cool problems where you can have a massive impact and legacy industries tend to be where there's a big footprint or install base.
Yeah, you all probably when you were doing Meraki, I think one of the most interesting things I had heard when I was in college and when Meraki was coming out is you just wanted to build networks that worked and that seemed like such an anomaly compared to others. Did you feel that internally too when you were building the company or how did that transpire?
Yeah. Again, like when we were doing networking, this is close to 20 years ago, networks existed and they did work, but you had to really be an expert. Like you had to be a CCIE or you have to be like a networking PhD to really understand how to build and manage a large scale network. So when we say networks that simply worked, it was really around how do we enable lots of people to build large scale networks and that was a challenge. That was the gap, right? And that was about configuration and provisioning and troubleshooting and all those kinds of things. So it's not so much that we were trying to fix what was broken. We were trying to do it in a different way that would open up the aperture and make it possible to build 10 times as many networks or our goal was like could you get a million of these networks deployed which eventually we got there. But to do that you had to really rethink the problem because you couldn't take the existing technology stack and push it into that scale deployment because you couldn't mint new network administrators who had the right training and sophistication.
And when you think about doing that with Samsara today is there like a similar playbook you think about on hardware software operations together or what are the parallels that you saw when starting Samsara?
So very different problem domains right like networking and Wi-Fi and switching and so on versus GPS tracking and dash cameras and asset trackers but the commonality is that if you provide an integrated solution the customer experience is radically different. It's fundamentally easier to deploy. It's easier to manage. It's easier to like make use of the data. Now, as we think about AI and I know you guys are doing a lot there with command, it gets a lot more interesting if it's all built together, right? And that has been a very common theme for us. It's a lot more work to be honest and you've experienced this, right? Like you have to get into all the details, but when you deliver it to the customer, it unlocks a lot more value and they can do much cooler things with it. So we kind of have always viewed that as being worth it. But it requires like really ingesting the whole stack. Really from the atoms on the hardware all the way up through the APIs.
You're one of the few CEOs that I think actively still is involved in like engineering.
Well, I think a lot of CEOs these days are spending time on product, but I think it's always just been natural curiosity and for me personally, I like spending time in the field with customers. I like coming back to the office sharing my notes. I write up these long emails with like kind of everybody that I observe and it's fun to be able to connect the dots and connect engineers to those problems. And you know the pace of technology innovation today whether it's on the hardware and silicon side or you know what's now possible with AI it's just so exciting like there's so many new things you can build it's fun to be able to brainstorm. So bringing the problem context to a bright group is always energizing.
What are you seeing these days at Samsara and other technologies you look at how it can impact infrastructure because I think again it's so different than doing in software so different in doing in other things but infrastructure you know you have to be more reliable it's more long-term how do you bridge that gap between doing all these new things but it's infrastructure at the same time?
So I think the first thing is to understand what are the P0s, non-negotiable like if you think about it like plumbing like the water has to flow and network infrastructure has become just extremely mission critical. It's not like a nice to have but you have people depending on it but now you also have IoT devices and other infrastructure sitting on top of network. So we have always tried to engineer the systems to be fundamentally reliable. So like how do you separate out the data plane and the control plane and the management layer, how do you make it recover and fail gracefully, how do you debug the system? These are all kind of real world problems. They become very complex but if you can solve them customers really appreciate it because they're running mission critical infrastructure and so today what we do at Samsara is we serve a lot of these infrastructure companies the stuff just has to work so you have to do a lot more software engineering to make it really recoverable but if you do it right you get credit for it.
And maybe personally or even all the tools you're looking at infrastructure is there something that excites you the most on what's possible and what can happen?
Well, I feel like you just blink and there's a new foundation model. Today is like Gemini 3 days, right? But yeah, I'm super excited about like the power of these AI models, their ability to reason and connecting it back to operations and infrastructure. You can now just ingest massive amounts of operational data, right? You can find hiccups, you can debug user experiences, you can stand up APIs, you can really connect these systems together in ways that weren't possible. So really excited about that. That's at the very top of the stack and down at the bottom, you know, silicon just continuing to advance at incredible rates. So what's cool is how much value you can get and how much you can deploy to the edge in ways that would have been like you know this only works in a data center. Now you can get it all the way out to like an IDF somewhere.
Right. And are there ways that you know you're working at Samsara differently today than you were a few years ago because of everything that's going on or how are you guys changing what you do?
Yeah, I think every business is rethinking how they use technology and like we're no different. We deploy a lot of AI agents and chatbots and things like that both internally and externally. So externally it's in our product similar to how you have it in the Meraki dashboard. Internally we have basically AI bots that our sales team can interact with or marketing team can use and so on. So we've been getting a lot of great leverage there. And then I think we're just trying to iterate fast right and to do that you need that shared context but you also need the ability to push stuff out. So moving to this kind of more staged or rapid deployment model where the fundamentals don't change and like rock solid, but then if you want to try a new software feature, you can roll it out to a couple thousand customers, get their feedback. And if you can be nimble that way, it really unlocks a lot of ability to brainstorm with your customers.
Yeah. Yeah, because I think the reason I was asking is I think the hallmark of the products you've built is they sort of feel like relentlessly easy to use at the top, but there's incredible complexity underneath.
Yeah. And I think you have to have an appreciation for that. And for us even from the early days when we were at MIT we were trying to factor out the complexity like let us take it on and that was made possible by you know the amount of compute you could throw at things it was fundamentally changing it changes like every few years. And so now as you can pull the complexity out you can kind of hide it from the customer the end user in a way that you're taking that burden off of them. So that's kind of been the belief right it's manifested in a few different ways over the years but I see you guys doing a lot of it too. I saw the hardware out there. Beautiful looking hardware on the outside. You crack it open, you're like, 'Okay, there's like a 12 layer circuit board. It's a really dense silicon, right? It's very cool.'
But, you know, I think one of the things that happens is like engineers that are customers and others are like, 'Well, you're taking that away from me, that complexity that I was part of. What's the right way to think about that?' Like over time, you know, you're used to doing things a certain way. Yes, there's better ways to do it with hardware and software, but it feels like encroaching a little bit in what I'm doing day to day.
Yeah. I think of it as it's really a question of what's the problem you're solving. So you do need to have sophistication in the product. So you can do things like configure VLAN tags and make it dynamic and do funky cool setups but that doesn't mean it needs to be done in a command line interface because actually the command line interface might get you into trouble. It might be brittle. The configuration might not roll out well. It may not recover well. So that's kind of the difference is like well what's the problem you're trying to solve? If you need that configurability, let's build it into the product, especially if it's going to benefit other people. I drive a Tesla, I use an iPhone. Like these are examples of pretty simple products on the outside that are incredible pieces of technology on the inside. So I think consumers have gotten more used to it. And if you think about it, every one of us were a consumer, right? And so it's retrained all of us to think there's a way now to get tremendous power with a really nice elegant interface.
