The Stove Guy: Sam D'Amico Shows New AI Cooking Features on America's Most Powerful Stove at Impulse

We scallop maxed.

Yeah.

Thank you, Claude, for telling me the temperature.

As you were.

Useful for everything from deciding where air defense sites are in Iran to scallops, apparently.

Thank you for having us, Sam. This is a really cool place. Do you wanna show us around?

Yeah, that sounds good. Yeah.

What is this?

So-

This is the-

... this is the Impulse cooktop. Um-

Mm-hmm

... it is, to my knowledge, the highest performance stove on the market.

Wow.

Uh, but, uh, a couple things here that's interesting. So one is we've, uh, we've set up the station with a ventilation hood as well-

Mm-hmm

... um, to kind of show you how much power this thing's got available.

Yeah.

Notice that the ventilation hood actually turns on automatically-

Wow, yeah

... when we start boiling. But look at how fast this is going. So we're putting 10-

Yeah, it's already going

... we're, we're putting 10,000 watts of power into a, like, m- medium-sized pan-

Yeah

... or pot. And, uh, we can boil about a liter of cold water about 40 seconds.

Mm-hmm.

This is, I think, a little more than a liter. Um-

Yeah

... but you can see it's already starting to boil.

That's already boiling. Man.

Um, and we actually... One thing to note here is we have a partnership with Zephyr.

Mm.

Um, so in stores nationwide starting relatively soon, um, possibly, like, at the same time this video's posted-

Wow

... um, there's gonna... you're gonna be able to buy-

It's going.

I'm gonna turn it down a little bit. You'll be able to buy a Zephyr-branded, Powered by Impulse products alongside the ventilation hoods.

Wow.

Now, and then we're also previewing features we'll have in the future, which is, like, coordination between the hood and the stove automatically.

Yeah.

So like I do that, this spins up.

Yeah, the speed goes to five, to two. Yeah, that's so cool.

And you can also... Why don't you play with it a little bit-

Yeah

... just to see how r- real-time it feels.

And these are magnetic, right?

They're magnetic.

So you just can-

So you pull it off, it turns it off, actually.

Wow. Oh, wow, it turns off as well.

So you have to click it to-

Click it

... do it.

Bet.

And then...

Wow, that is so cool. And it's so responsive.

Yes.

It's, like, very-

We want it to feel analog, just like how a gas stove feels.

Mm.

Because, like, most other induction stoves can't do this.

Yeah, it takes a lot of time.

Um, we had to redo the power electronics and control systems from scratch, and we control all the s- the firmware and hardware and everything like that to actually make this experience possible. So everything from the turn of the knob to the bubbles being formed there is under our, our full control.

Wow. Yeah, that's amazing.

There's a lot of other things here. Is this testing? I see a lot of pots, pans-

Yeah, yeah, yeah

... a pan.

So when we get interesting contacts from customers for, like, "My... This pan works a little funky"-

Yeah

... or, like, "Does it work with X?" We end up just going and buying stuff and testing it. We also have a bunch of pots here that are actually wired up with thermocouples so we can actually just reliably test, like, our temperature control, power control alg- and power control algorithms so that, like, we know that there's no regressions when we go and release stuff to customers.

We've actually... We're in the process of qualifying an update that, like, triples the performance of our temperature control system.

Wow.

Which means... When I say that, it's like, what does that exactly mean? It means it gets to the temperature you set about-

Mm

... three times faster than it currently does.

Wow, that's, that's very impressive. And I guess for QA, do you guys have, like, a system where you have a lot of people come in and test out stuff before it goes out, or how does that work?

So we have, um, we have... We- we're, we're a small team. We're, like, 30 people total.

Yeah.

Um, the... We QA every build we send out, uh, to customers, obviously. Um, and then what we also do is m- many of us have units at home, and so that's kind of the first wave. So it's like people are cooking at home, and they'll see...

you know, they'll see the builds. But obviously, like, for, like, the baseline, like, let's, let's test every edge case, we do it basically here-

Wow

... on this station.

Wow, that's, like, the true definition of dogfooding. You're, like, trying it out yourself and making food.

Yeah, yeah, yeah. But hopefully you're not eating dog food. But...

That's true, that's true. Here we have the parts.

Yeah, so we basically... Yeah, here's, like, kind of the stove exploded on a table. But basically-

Ooh

... we went and, like... The idea here is, like, we went and effectively reinvented the stove from first principles around the idea of having a lithium iron battery at the center of the system. So what does that exactly mean, and how does that work?

So basically, we put a three kilowatt hour LFP battery, which is lithium iron phosphate, inside home appliances, starting with stoves. That gives us an additional 15,000 watts of additional power. So that's, like, double what a typical induction cooktop has in terms of aggregate power across all the burners and stuff like that.

Mm-hmm.

On top of what the wall supplies through a plug.

Mm.

And so we're able to actually bring to bear significant performance improvements over other appliances just from having more watts, watts available. So what we then did is we were like, "Okay, let's reinvent everything upstream of that." So we went and built our own custom power electronics and control system.

So, like, this is one of our induction driver boards.

Yeah.

Um, we've also got, like, this temperature sensor technology, which we'll be using later when we cook.

Yeah.

Um, we bu- When we... And then our induction coil, we actually ended up designing all of this from scratch.

Oh, wow.

Like, we did all the magnetic simulation, et cetera, ourselves. And this allows us to actually control everything from the turn of the knob to the pan getting hot.

Mm-hmm.

And we control that entire end-to-end flow with kind of our own firmware, embedded systems, and, and high-level software.

Oh, that's so cool.

Um, that has allowed us to basically build the highest performance stove on the market.

Mm-hmm.

Um, now why do we have all this stuff splayed out? Well, I also sell this to other major appliance OEMs. So basically we built a modular platform, we call it Impulse Core. Um, and if you use Impulse Core, you actually have the hi- You will get the highest performance applianc- You'll be able to build the highest performance appliances yourself as well.

Wow.

So we've got multiple deals with multiple major appliance manufacturers. Some of which have announced, some of which have not announced. And we'll be shipping globally this year, uh, through a number of partners.

Wow, that's, that's exciting. So you have the main product that you guys build entirely yourself, and then you also have the core product that you sell out to-

Exactly. And that gets us channel access to, like, every form factor of appliance-

Mm

... every market we may not be into. And also, like, it gets us into stores across the United States without us having to have, like, a big sales team. So we can basically operate much more like a software in, like a software company-

Yeah

... than, like, a hardware company that needs a giant logistics footprint, and ships large cardboard boxes across the country, and has installation services, and all this other stuff. We can really be kind of like- ... intellectually heavyweight, but a logistics lightweight, basically.

Yeah.

Yeah.

How long does it usually take for you to build something? 'Cause you even said from the start thinking through first principle. It's kind of reminding me of how Elon thought through, I think, like, SpaceX-

Yep

... as well, and I assume there's a lot of inspiration there. So I guess in terms of building these things out, iterating through, like, what are the timelines looking like?

Yeah, so it took us four years flat from receiving... Or wait, no, three y- it was a, it was, it was, it was 2021 we started the company.

Mm-hmm.

In July 2021. We shipped our first revenueable units, I believe the first one went to the American Housing Corporation guys-

Wow

... in, like, June 2022, 2024.

Mm-hmm.

Or 2025.

Yeah.

So this was, uh, for three, three years and 11 months later was, was how long it took end to end to do the entire thing from, like, company formation, like, literally I filled out Stripe Atlas- ... to, uh, to landing products in people's homes.

Wow.

Um, I think we could've probably shaved six months off of that-

Right

... for various reasons. But, like, you know, typically building a product where end to end you need to basically reinvent, like, core modules.

Yeah, every single part.

And there was also a heavy compliance requirement, so basically getting this through the UL 858 tests, getting this through, um, some of the UL, like, other UL requirements we had to, to meet. And also, you know, there's a bunch of...

And we had to build, like, many, many, many units, like 200 plus units, to get... Before we even shipped one to a customer, and put them through a battery of tests to make sure they were reliable, safe, et cetera.

Mm-hmm.

And then we also had to develop all the software that runs on the device, too. So it's not just you build the thing, but you have to build the thing and then make sure that the software on it is good.

Yeah, that works and-

Yeah

... continues to get updated.

Yeah. Exa- exactly.

Mm.

So, so basically, typically the length of time it takes to build a complex hardware device, I, I think it's like, it's two years if you don't have to really invent, like, new modules and systems.

Yeah.

It's probably three if you, you know, if you do.

Yeah.

Um, just based on past experience from, like, my backgrounds in consumer electronics. Worked on everything from Google Glass to Oculus, you know, other things like that. So worked on mostly face computers before I jumped into this space.

Yeah. No, that was one thing as well, is starting consumer hardware and then pivoting now to... It's s- like, a passion as well. 'Cause I heard that it was kind of prompted with a pizza trip to Japan.

Yeah.

Going well with it.

It was r- it was actually not... It was a- well, I was not in Japan for pizza. I was in Japan for the IEEE Haptics World-

Mm-hmm. Okay

... conference. So we were working on game controllers at the time.

Gotcha.

So if you've used the Meta Quest Pro, I was the controller... I, I, I, I came up with that controller program.

Yeah.

Um, back in, like, 2018 or 2019 or something like that.

Mm-hmm.

Um, but, um, I was, I was in... Or in 2019, I was at the IEEE Haptics Conference. I was like, we went to this really good pizza place called Savoy. I was like, I got my pizza back in, like, 30 or 45 seconds, and I'm like, "What is the power density required- ...

to, like, cook a pizza that fast?"

Yeah.

And I was like, "Well, that's, that's a lot of watts, and you can't get that out of, like, a tabletop toaster oven or something like that."

Yeah.

How could you build a tabletop device with that level of power? And it's, like, the answer is you slam a battery into it.

That's great. Yeah.

And if you slam a battery into it, you're now not... You c- you've decoupled the energy input from the wall with the device's power outputs. You can decouple the user experience from the grid.

Wow.

And that level of, like, approach has not been done in kind of the major home appliance space-

Mm-hmm

... at all.

Mm-hmm.

And that kind of gave me the spark of an idea. Um, I also realized another thread, which was I was, I was, I was working at, I was working at Meta at the time, or I guess Facebook. It was called Facebook back then.

Yeah.

Um, and realized that if I wanted to build my own hardware company, um, I was kind of, like... I think you can look at, like, Palmer Luckey with Oculus, and it was like, how did he prototype that and how did he figure out, like, where was, like, the, the, like, where could you move things kind of, like, revolutionarily a step forward?

Yeah.

And it turns out he could do this, like, while, like, I think living in a trailer in his parents' driveway with, like, meager resources because it was just, it turns out that, like, the low-hanging fruit had not been harvested yet.

Right.

And the low-hanging fruit was like, oh, phone displays are really good. You can make a VR headset with a phone display or a tablet display.

Mm-hmm.

And the optics for this were, the optics for it were achievable by someone who, again, was like, you know, basically, ba- like, he, he was, he, like, basically like a college age individual without, like, crazy resources, right?

Mm-hmm. Mm-hmm.

And so, and so that said, you had to build a real company and do all this other stuff, which he then did. But, like, the, the big part of this was

you did not have to be a large technology company to move the needle forward on the technol- on the, on the, on the capabilities.

Yeah.

Like, typically to design something, like, oh, I built, like, Face ID at Apple. It's, like, a multi-hundred million dollar program where they had to acquire, like, an Israeli computer vision company-

Yeah

... that had the hardware. Like, like, one of my friends worked on that, and it's like these are, like, big industrial efforts to kind of move the ball on, like, making the iPhone better, for instance.

Mm-hmm.

And that's even, that was true 10 years ago. It's even more true now.

Mm-hmm.

Um, and so I'm kind of like, well, where can I do this? Like, obviously VR got harvested and-

Yeah

... I was, I, and I was seeing even in VR it's like if a... Let's imagine I went, left Oculus and I went to start my own VR company, which I could have done.

Yeah.

Um, well, the first thing is, well, you need to spend a lot of money on component engineering and developing new components-

Mm-hmm

... and stuff like that, and it's like, it's not a reasonable ask for a VC to be like, "Yeah, give me $100 million to make custom displays." Like, you know, custom displays, custom chips, custom cameras. Like, all of this stuff has to be custom, custom, custom because-

Yeah

... well, you're up against the Vision Pro, and they're spending-

A lot

... a lot as well.

Mm-hmm.

There's no, like, there's no, like, architectural insight that makes, that you can be more cl- you can outmaneuver them by being, being more clever.

Yeah. Agree. Super agree.

And so, and so the second thing it was, was, well, friends with a lot of folks who kind of gave me this feedback on the consumer electronic side, which is like, I want to build a company that didn't, doesn't have to get acquired.

Mm.

Like, the big, the big goal here is, like, if you need to have an app store to make the thing work-

Yeah

... well, you need to subsidize all these developers to basically come to your platform.

Mm-hmm.

And I want, uh, you don't need an app store for home appliances. Now, maybe some people think that, that that would be a good idea, but, like, ultimately I do not believe you need an app store for home appliances.

That's true.

And so, like, you can ship them a stove and as long as it cooks, you are good.

Mm-hmm.

Meanwhile, if I ship someone a game console- Well, unless I'm paying game developers hundreds of millions of dollars, you know, aggregate, probably billions in some sense if you look at how much, like, Meta Reality Labs spends on content for the Quest headsets and stuff like that, it's like, that's an, it's an expensive proposition to actually maintain that two-sided marketplace-

Yeah

... especially before the user base is mature.

Mm-hmm.

Um, I didn't wanna have to run into problems like that, basically. So I was like, okay, two things. Let's pick on a low-hanging fruit space where no, none of the big tech companies are playing right now, and two, let's, uh, make sure there's no, like, app store or ecosystem requirement.

Make sure, like, I could literally ship one of these and it's good.

Yeah.

Um-

But the value is direct.

Yeah, the value is direct.

Mm-hmm.

And so, and so those are kind of the, those were kind of two key guiding principles, and obviously realized that home appliances were an awesome space to get into because, well, there's a secret to getting better performance out of it.

Yeah.

And then it was like, okay, well, what do you do first? And so the pizza oven thing was kind of a cool idea, but I'm like, well, everyone's got a stove, not everyone has a pizza oven.

Yeah.

And I'm using my gas stove at home and I like my gas- I like my gas stove at home, and I had tried a bunch of electric stoves. I tried, I had an induction hot plate, I had all these things, and I'm like, these things are all...

Like, no one really got the UX right or any of these other things. And so I went and basically studied the space pretty deeply and realized that, like, yeah, this is probably the right entry point because it turns out the stove is...

If, if you look at someone's home and you put an electric stove in someone's house, it's like 60 amps of their, like, 100 to 200 amp panel capacity is, like, just an induction stove, which is crazy.

Wow. Yeah.

So if you can make a better one-

Mm-hmm

... because you've got even more power available, and you can reduce the requirement that it needs to, like, run on the ho- the home.

Yeah.

You can then install induction stoves in every house in America, and you can install better stoves in every house in America than anyone else.

Yeah, much better stoves.

Much better stoves.

Yeah.

And then the last piece is, like, well, now I've got a battery in everyone's house.

Mm-hmm.

And if I have a battery in everyone's house, there's now energy storage all over the grid, and, uh, we all know, like, there's kind of a, the, there's both a power generation concern, but also-

Yeah

... a congestion concern on the grid where, like, we're adding all of this new load growth. It's everything from AI to people getting EVs to also people's homes electrifying as well.

Yeah.

Like, they're getting heat pumps and air conditioning and stuff like that. And so if we can basically put energy storage on the grid kind of, like, invisibly in these, like, middle, midsize machines, we're gonna be in a very, you know, much better spot than if we didn't do that.

So that's been kind of my, like, thinking. And then there's, there's one last piece-

Yeah

... which is I think worth raising, which is we build this thing like it's a phone, and it's not... Like, it's ho- We view this as a consumer electronics device. We don't really build it like it's a home appliance, and there's a philosophical difference between what I would call kind of like electricity 1.0 and, like, what I would say is, like, the modern electric technology stack.

So it's like batteries, power electronics, embedded compute, electromagnetics, et cetera. Like, everything is getting built around a very relatively narrow set of technologies. So like, okay, we've got a lithium ion battery.

Yeah.

You've got, like, embedded computer stuff. You've got motors or induction coils or whatever.

Coils.

And then you've got power electronics. Um, and the real advantage of this is this is allows you to build software defined hardware products.

Mm.

And so if you think of, like, the big innovation of, like, Tesla versus other car companies-

Mm

... like, be- like, Tesla did a couple key things. One is they-

Yeah

... they made the electric car work. But then the second thing they did is they built it like a phone-

Mm

... where they controlled... Like, and the Model S was actually... Like, I would, I would argue the, the Roadster wasn't exactly this, but the Model S really was this. It's like this was a car designed end-to-end like a consumer electronics program where everything, including all the modules that go into it, were coherent with the original vision-

Mm

... versus, "Oh, we have to go use this component from this vendor," or whatever. And because they did it that way, it's just a better car than everything else. And obviously, our friends across the Pacific in China figured out exactly the same strategy, and they're really good at consumer electronics-

Mm

... so now they've got some of the best cars in the world with, like, Xiaomi and all this other stuff.

Yeah.

And I think that's a bigger lesson than, say, like, Elon talking about vertical integration or anything like that. It's like there's a consumer electronics mindset of, like, you need to own the end-to-end product all the way through.

Mm-hmm.

And, like, my view is, like, everything from this knob turning-

Yeah

... to this coil heating up a pan, we control, and making sure that we do that allows us... We, and we have full software control, which allows us to then, you know, do more things beyond what we're shipping on day zero as well.

Yeah, and it'll be significantly better than anything on the market.

And it's, and it ends up being a platform for, like... And you can see with, like, Tesla with FSD.

Mm.

You can see Tesla, like, literally going and they had a video where they basically went and said, like, the, uh, like the, they were sunsetting the Model S and Model X, and they basically will go and show that all of those technologies are going into Optimus.

Mm.

And so getting good at the technology stack is transferable beyond that, basically.

Mm-hmm.

It's very evident that you put a lot of care and quality into even just every tiny piece of the hardware, but one kind of question that I assume a lot of audience members have as well is do you have any plans to release a single cooktop, I guess for-

Oh, everyone asks for the single burner

... yeah, really

Everyone asks for the single burner.

Yeah.

I think we're gonna have... We're, we're working on what, what, like, tabletop devices we're gonna, we're gonna ship, you know-

Yeah

... in the, in the, in the near future. I think if you're, if you're interested in that or, like, a two burner, like we're, we're, we're definitely thinking of you guys, and I think that that's something that, uh, we're gonna try to have a solution for in the, the coming year or two or three.

Great. Thank you for showing us the tour and the space, and I hear this is your dev unit that you have pulled out for us to show some cool tricks.

Yeah. Usually we have, like, a, an actual production unit, but this is, like, a dev unit that I've been working w- w- with behind my desk, but we put it in the studio for, uh, today to show you some new stuff, so.

Awesome. Yeah. We could cook some scallops and some pasta, so do you wanna get us started?

Yeah. So I'm an idiot, and I don't actually know how to use a stove, so let me, let me... Let's see if this, let's see if I can, like, ask the stove what to do.

Set the front right burner to the right temperature to sear scallops at.

This is AGI right here.

AGI.

If it gets it done.

We got it. We got it.

Sear the scallops

Yeah, we'll see if it work-

... at the right temperature.

We'll see if it works.

Set the front right burner- Oh, wow

Cool. So it'll set to- So it's gonna set some high temperature to sear scallops.

Wow.

I don't know, I tr- I tr- I tr-

And automated everything as well

... I trust, I trust Claude. I don't know, we'll see what happens.

Wait, so Claude is the model that's running behind this?

Yeah, it's running, it's, it, yeah-

Okay

... it's, it's running Claude. I need to... I probably should run some, like, s- super fast Cerebras thing, but we'll get to that later.

Yeah.

Get to that later. Um, this was also part of an experience for me to teach myself on how to build agents end to end.

Yeah.

So basically I was like, well, the world's different, I need to actually learn how to do this stuff.

Mm-hmm.

Um, even though, like, I'm a hardware guy, I do have software background, I did take software classes in college. Um, I did... I've deployed software in production facilities and stuff like that, mostly for servicing hardware companies, but-

Yeah

... let's try to build some AI stuff, so I went and put Claude on the stove.

Yeah. No, that's very cool. I've never seen a use case of Claude in the kitchen, in some sense. I guess, how has AI also helped in development, or has it also accelerated, whether it be R&D or testing things or trying out new products as well?

Yeah. So I started... So I've been using the AI dev tools since-

Yeah

... they basically came... Or I've been using AI for software development since it basically started being acceptable at it.

Yeah.

So, like, I was... I, I... When Sonnet 3.5 came out, I was like, okay, this is, like, we're actually... This is good enough to start. Like, you could vibe... You can actually vibe code.

Yeah.

But basically I was like, I was playing around with Cursor.

Mm-hmm.

I was, I was actually also just using the web, web, web UIs, and just, like, being like, "Give me a website I can download."

Yeah.

And I was, like, pretty reasonably impressed that this stuff could actually work.

Mm-hmm.

Um, which is cool. But, uh, when Claude Code came out, it was like, it was like... Well, when I had access to Claude Code, it was, like, April or something like that.

Yeah.

April or May th- this past year. I was like, okay, shit, this is actually, this is actually going for real, right?

Yeah.

And what I started doing is I was like, I noticed a gap in what we had for the company, which was we didn't have, like, fleet telemetry that was, like, our, under our own control. So I, I was like, okay, really important to actually know what's going...

If you're about to ship a product, it's really important to know what's going on with your product to make sure that, like, you know, make sure customers are happy, but also, like, if something goes wrong, like, you can go look it up by, like, seeing, oh, there's a bunch of error flags in this unit or whatever.

Yeah.

So I started using, I think this was Sonnet 4, I started using Sonnet 4 with Claude Code, and I was basically went and built an end-to-end fleet telemetry system from, like, Terraform, like, like, all... Like, like, infrastructure in Terraform deployed on AWS.

I've never done any of that stuff before. And it was kind of this, like, you know, hardware guy learns enterprise SaaS sort of, sort of, sort of venture.

Yeah.

But it was also a, like, how to use the agentic tools to, like, you know, really, really build something durable.

Yeah.

So that's, that's how... That was, like... And you can look at, like, my GitHub commits and all that other stuff, and it just basically took off at that point.

Yeah. That's amazing. Wow.

So by the way, the temperature control on this thing is, like, it's already at temperature.

Yeah. No, this is perfect.

It's already at temperature and it goes.

And it's not, like, smoking or causing it to get too-

It's not smoking or anything like that.

But yeah, so you can also-

It's just raw dogging these scallops. Well, not raw, but you know- ... you get the idea.

Great. Yeah, we can do, like, four, four scallops, or maybe five. I feel like the temperature... 'Cause that's one thing, too, right? So this will probably retain the heat of the pan, um, even as you add stuff, right?

Well, that doesn't matter. So the, the, the thing is, w- the way this product works-

Yeah

... we control the temperature.

Mm-hmm.

So this thing is holding 446.

Yeah.

And you notice when I added that scallop, it starts dipping-

Yeah

... but it's gonna put more power in to kinda keep it under control.

Dang, so you have, like, complete control over-

Complete closed loop temperature control-

Wow. That's amazing

... end to end. So this thing, like, doesn't smoke, doesn't burn.

Yeah.

And you'll see, like, you can start seeing the crust on the bottom of the scallops right there. I don't know when we're supposed to flip these things.

Yeah, we could probably flip it, I mean, in, like, 10 more seconds, like get-

Okay

... more color. But yeah, that's great. And I assume you can... Wow, and it also shows, like, the temperature down-

It shows the temperature of, like-

... to the degree. Yeah

... you know, like, there. Yeah.

The burner status. Great. Okay, yeah, make sure-

Yeah, so I can go look at that.

Wow.

That's looking pretty good.

Yeah, that's looking really good.

Yeah, so I can go... I kind of seared that one on a weird angle, but you know, you get the idea.

Yeah, but these look really good. And I love how it shows the real time burner status-

Yeah

... to be able to understand, like, where it-

Yeah

... 'cause it also just dipped down, like, 10, 10 degrees-

Yep

... if it shows. Wow.

And also, you can, like, lift the pan up and it resumes, no problem.

Yeah.

So, like-

Mm

... pretty straightforward.

Yeah. No, that, that's amazing. And then I guess we can start with the water, right?

Yep.

'Cause it also boils water at an insane pace.

Yeah. So we can put up to 10,000 watts of power into any burner on this thing.

Yeah.

Um, so that's like three times what most induction stoves are, and, like, it's, like, five times what they are when they're not in that, like, boost mode-

Yeah

... in a bunch of cases, which is often time restricted, because they don't have the temperature sensor to know that it's, like, under control basically.

Yeah.

So I just put 10,000 watts into this thing.

Wow, it's already-

And you can see the bubbles are already forming-

That's insane

... on that thing.

That's pretty insane.

And it's like, okay, it's already, you know, it's already 150 degrees at the bottom. It's like, it's climbing really rapidly.

Great, and we could get the pasta over there. Do you wanna grab that?

Yeah.

And then we can start moving... That's a little bit... Great.

You can probably just dump it in, because honestly-

Yeah. Should we? It's already-

How, how much do we wanna put in?

We could do three of them, or...

Yeah, we just, we pasta max.

Yeah.

Let's do it. Let's do it.

Yeah. Honestly, yeah, let's, let's just do all of it. Great.

Yeah. Oh, yeah, I did, I didn't realize-

Let's get that

... we were at four. These are probably... Are these probably good?

Yeah, I think these are good. Yeah.

Yeah, yeah.

The color looks good. So we can... Here, do you wanna put them on this plate?

Yeah.

Wow, these look amazing. Wow, the color and everything.

We scallop maxed.

Yeah.

Thank you, Claude, for telling me the temperature.

Great.

Wow. That's amazing.

Useful for everything from deciding where air defense sites are in Iran to scallops apparently.

Yeah, the whole, the whole spectrum. Great.

Yeah.

Yeah, makes sense.

So we gonna use this pan for anything else, or are we good?

Yeah, so I think we're good. We'll just boil the pasta, and then we'll use this other pan to kinda assemble the final dish.

Okay. Sounds good.

Yeah.

So I'll go put this guy over there.

Yeah. So that's great.

And then you can... Yeah.

And then, yeah, we could put this over here for now.

And then how long of a ti- how, how much, uh, time do we, we need to wait for this thing?

Um, this one I think is only five minutes.

Okay.

But yeah, we can-

So this thing is gonna boil over pretty fast, but we can... We'll, we'll turn it down in a second.

Yeah. Great.

So you see how... Look at how crazy this is, by the way.

Yeah, it's-

And look at the control authority. It's like, oh, basically nothing.

Wow.

Oop.

Yeah, it's instant. It's like-

So let's set it to, like, two. This is, this is what high is on, you know-

Mm

... for mortal, for mere mortals. We'll just leave it like that. Um-

2.2

... and then

set timer for five minutes from now

See if it works, I don't know.

Yeah, wait, this test unit, I mean, it's already put in a lot of work.

Yes, we got a timer over there, it's for five minutes.

Wow. Yeah, and you can see the temp, the time, everything.

Yeah.

Wow, that, that's amazing. I guess when you go through iterating through ideas, is it just a lot of dogfooding and then you figure out which one-

Yeah, and also, it's also, it's the fact if you can use AI now to prototype stuff just so much faster.

Yeah.

Like, this UI is a concept UI that we, um, our, our, our, our product manager worked with her design firm on, and she's already, like, two steps beyond this.

Yeah.

But basically what we did is I was like, "Okay, uh, just give me the latest Figma," and I took the latest Figma, using the Figma MCP-

Mm

... and managed to get this running on here.

Wow.

And this UI was enough to basically give us the starting point to be able to build some of these AI features so you could see stuff.

Wow.

So, like, I, I'll give... While this is going, we got, you know, we got four more minutes, so I can show you some other stuff while we're here.

Yeah.

So I can be like-

Let's see it

... pull up a recipe for Kenji López-Alt's Korean fried chicken.

And this is going to the internet, so it's gonna, you know, it takes, like, 30 seconds or something like that-

Yeah

... because, again, I'm like, I'm not, I'm not, you know, I'm not, like, LLM tool call optimizer. I'm, like, a hardware guy. But basically, but basically, goes, goes, does a web search, gets the URLs, goes to the various sites, figures out a good recipe that works, and then it, um-

Wow, yeah

... and then in the a- in the agent harness, it basically converts it to a, a defined schema-

Uh-huh

... which then shows up on the screen. So it goes into recipe steps and stuff like that. So then... And then you can use the knob you actually had to go do it. So by the way, it has-

Wow

... its own timers.

Yeah.

So it'll be like, "Coat the wings with whatever," and then it says, "Wait, you know, 30 minutes," and it'll, it'll go do that. And then when you get to preheat the oil, it actually sends a command to the burner.

So it'll actually say like, okay, it'll say, uh, like, it, there's a command to confirm, so I can click that.

Yeah.

It's gonna set that to 351/177 C. I- there's some bugs in terms of-

Mm

... unit use and stuff like that, but bear with me.

Yeah.

And then it's like, "Make the batter."

Mm-hmm.

Go do it.

Yeah. Let's wait for so they can see it.

And then, you know, fry the wings, and it'll actually have a timer while you do it. So it's like, okay, we'll set it to, you know, set it to the preheated temperature, do that, and then, you know, serve or whatever.

Yeah. We-

And so-

This UX is very nice where it even shows the image, lets you just-

Yeah, it shows all the stuff

... automatically set the temp.

Yeah. And it's super flexible.

Everything.

And it's super flexible, too. So like I'll, um... I can go pull up something else. It'll be like, um... Let me find another thing. Um, pull up a recipe for egg fried rice, but modify it to include char siu pork.

We'll see if this works. I don't know. So char siu pork-

Pull up

... egg fried rice. So it's giving you a bunch of instructions. So it says like, "We don't have day-old rice, unfortunately, so we can't cook it right now."

Yeah.

Um, but it goes to scramble the eggs, sets the temperature to the scrambled eggs.

Mm-hmm. Wow.

And then it's like, "Get the char siu pork." Apparently you pre-marinate it. Assumes I have char siu pork on- But you can get that on Oui, actually.

Yeah.

Um, and then it actually raises-

If you want pork

... the temperature up, raises it to 200.

Wow, yeah, I can see it going.

And then, and then it lowers it for the aromatics.

Wow.

And then it's like stir-fry the rice, and then finish and serve. And so you can go... What's fun is even scrolling backwards, it makes me do that, but that's just a, that's just my own limitations and my own- ...

agent harness.

Yeah.

But you can see that this actually, like, this actually works, right?

Yeah.

So, um, also this was actually commanding this one to this temperature to do it, so it can do multi-burner stuff.

Wow.

So it was basically saying like, "This is a wok."

Yeah, you can use multiple.

"Heat the wok over here." So when it was ba- ba... I forgot to mention that. So it was doing this for all of the, like, prep stuff.

Mm-hmm.

And then it was basically saying, "Assuming there's me a wok here," that would be going and doing it.

Yeah.

And that set that to 350-

Wow, that's very smart-

... for, for the tossing

... so you can do multiple things.

And it'll do multiple burners and knows, it knows the capabilities of the stove.

Wow. And do you wanna start the, the burner? We could do-

Yeah, what are we s- which we, what are we, what are we doing on this one?

Um, so it, we just need to turn it on medium, medium-high, um, ish. I don't really know the... We're just bur- um, melting some butter, rather.

Oh, you just wanna melt some butter.

And then-

Okay. Yeah.

Yeah.

So you know we can melt butter... You wanna brown the butter or no?

No. Um, but we-

Okay, so you won- you don't wanna brown the butter. Cool.

Yeah.

So I'll set it to 250-

Mm-hmm

... or 260, which will not brown the butter.

Great.

Um, and then should we just-

So yeah

... get some butter?

Some butter. We have some butter here.

How much butter do I actually need?

You could... We can cut it. Do you wanna cut, like, a third? You could put it down the table.

Yeah.

Just there on the table.

That sounds like a, that sounds like a, a patriotic amount of butter.

And then we can, yeah, let the butter go.

Yeah.

And then-

So by the way, this will, like, this will just melt, and it won't burn because of the temperature control. Okay.

We could just dump or, like, as much as-

How much we want?

So we could do, like, half of that. But if you wanna make it more cheesy and, um-

We c-

... rich, we can just do everything.

Okay. I mean, I don't think we need to do everything.

Yeah.

I think cheese maxing is, you know-

Cheese max

... good to a certain point.

Yeah.

Also, if you cheese max, you're not looks maxing.

Indeed. That's true, 'cause, you know-

But this cheese-

... train of thought

... this cheese and butter are gonna MOG, as we say.

Yeah.

Shout out to our brain rot overlords. Um...

Yeah.

So this is... By the way, this is just gonna melt. It's gonna be awesome.

Mm-hmm.

So-

Yeah, it's already boiling

... I don't know what the melting point of Parmesan is, but, you know, it's, we're probably above it.

Mm-hmm.

Um, and also you'll notice that nothing's burning.

Yeah.

Nothing's brown.

Which is very... And it's also not sticking as well, like the cheese.

Yeah.

So-

Because this is basically enabled by this active temperature control technology.

Yeah. No, it's, it's going well.

Um-

Do we wanna add a little more butter, and then we can...

How much more? Probably, like, that much more.

Yeah.

Yeah.

Probably, yeah, like, another chunk. Perfect.

Cool.

Great. Wow, everything is moving just very in a controlled manner, which is very nice.

It's very chill.

Yeah. Very, very chill.

And that's the idea here is, like-

Yeah

... it's like, it's like you can... This takes a lot of... I mean, you can make... You can overcook things, but you can't, you won't be able to burn things because-

Mm-hmm

... of the temperature control.

Yeah, that makes sense. Great. Okay. And then we could probably-

Feels like it... Look, I, I think we want more cheese.

Yeah.

I, I think-

I would agree. Yeah.

I think-

We could add a little more cheese.

Yeah, I think we wanna, you know... Yeah. Gotta offset those Chinese peptides in our systems, you know? Give them more temperature. It's good.

Indeed. Yeah. I guess while we're going, do you also have call-outs that you wanna make? Are hiring, you know, new launches, different things?

Um, I mean, I think we're just really excited, we're just really excited to, uh, you know, get the product out. We're la- we're launching in showrooms nationwide with Zephyr, like-

Wow

... probably about the time this video's out.

Cool.

Um, they're, they're, the... I think they're, like, showrooms getting their units right now. So basically we're really excited to be In hundreds of store... like, literally hundreds of stores nationwide.

Wow.

And I honestly think that there's this, like, story here of, like, what are we actually building?

Mm-hmm.

We, uh, we realized that

home appliances, we stop calling robots robots when they work. And-

Mm

... we figured out how to automate some of the hardest problems in the kitchen.

Mm.

Like, holding temperature consistently. Like, the fact that, like, this just works.

Yeah.

And the fact that, like, I can... I don't have to, uh... Like, there's no browning, burning, anything like that.

Yeah.

Minimal sticking. Maybe a little bit, but yeah.

Yeah.

You know.

We can add some.

Before anything like that.

Yeah.

Yeah, add the pasta to that.

But basically-

Mm-hmm

... we figured out a lot of the key automations. We can actually... And then we built the software-defined hardware product-

Yeah

... that allows us to layer intelligence on.

Mm-hmm.

But then the question is, like, well, how are you gonna get these things into people's homes for real?

Mm.

And I made a realization a while back that's basically like, well, you're not gonna alter human behavior of, like, how they get machines in their kitchen.

Yeah.

Like, you need to show up where they're actually going and buying these things.

Mm-hmm.

And so the realization was, like, we've got this deal with Zephyr, who makes, you know... They're one of the top, or the, the top ventilation hood brand in the United States.

Yeah.

Um, well, beneath every hood there's a stove.

Yeah.

And if we now are in all of these countless, you know, stores across America, um, this is gonna be a lot of Americans' first contact with, like, products like this.

Yeah.

And-

It's true

... we can then deliver experiences, you know, empowered by AI.

Mm-hmm.

Like, as the... The stove is, like, an embodied intelligent thing.

Yeah.

And it's a logical first step into a broader spectrum of, like, bringing, like, actual smarts to the home for the first time. Like, I think the first era of smart home was, like, we can actually put phone chips in products.

Yeah.

The next era is like, no, these things are intelligent and useful, and they do things that are, like, not a party trick, but actually, like, hey, we looked up the temperature to sear those scallops today.

Yeah, that's true.

And it just worked.

Yeah. No, that's great.

So-

I, I l- love the mission. I think, you know, having it in households worldwide eventually-

Yeah.

Mm

So we're actually l- we're also launching globally this year as well.

Mm-hmm.

We've got a really exciting set of announcements queued up for that as well, but more to sh- more to come on that side.

Great.

Um, this is looking fire.

Yeah. Should we add the scallops in?

Yep.

And then we can then, like, a final mix around, and we should be good. We could try it out. But yeah, thank you again for showing us the test unit and just the amazing-

Yeah, so this is, by the way, this, this unit is just a slightly earlier manufactured unit.

Okay.

But it happens to be, it has, happens to be unlocked.

Mm.

Have an unlocked bootloader. So I've been, I've actually been realizing that, like, I need to step up my game. Like, I mean, I think I heard, like, Sergey Brin's coding now. You know, it's like-

Yeah

... something like that. So, like, if you're a 30-person st- hardware startup, like, the CEO better be coding, right? So.

That's true, that's true.

So that's, that's been-

Founder mode.

Yeah. And, you know, as long as I don't get one-shotted by GPT or, you know, other stuff. I avoided the whole 4-O, you know, thing, but yeah.

Yeah, that's fair. That's fair. No, but it seems like you're doing pretty cool things. And I guess for the audience, will these AI tools as well be out when the, uh, stoves are available?

Yeah, so the goal is that with... If you buy an Impulse cooktop today, these, these features are... This is a preview of features we can do.

Okay.

Um, so no guarantees that the features are exactly parity.

Mm-hmm.

But, uh, I wanna have these features out this year.

Gotcha.

In, in, in, in homes across America.

Great. And it'll probably be, you know, similar to how Tesla does software updates.

Yep.

Like, you'll continue to do software updates.

Yep.

So...

We, we've done, like, over 10 OTA updates, um, so far for customers.

Sweet.

Um, we've got one coming up probably by the time this video's out that, uh-

Mm-hmm

... dramatically improves temperature control.

Great.

Um, so you didn't notice it before, but basically it, it's signifi- it takes about 40 seconds to get a pan up to temperature.

Mm-hmm.

Um, on the public release right now, it's about a minute 50.

Wow. Yeah.

Um, but we're really excited to get, you know, get kind of the performance that you see right here-

Mm-hmm

... um, out to customers, uh, very soon actually.

Great. Yeah, that's exciting. I feel like a lot of home cooks-

It's going through final QA right now.

Mm-hmm. Great. We can try the pasta now.

Also notice the butter is still not burned.

Yeah. Don't see any brown butter. This is great.

There's a bunch of, like, little, little bit of, uh-

Yeah, Parmesan

... little bit of little Parmesan, uh...

We have some Parmesan crisp though.

It's probably gonna taste pretty good, honestly.

Yeah.

Yeah, that's fire.

We can try a little, and then we can have some also... We'll have some judges as well trying it. Seems very cheesy. I... We didn't really try it for salt, but I don't know.

I think it's okay.

Mm.

I think you put enough salt in the water.

Mm-hmm. Mm. Yeah. I'm a very big Parmesan fan, so I, I agree.

I think this is good.

Yeah.

Yeah. I'm gonna, I'm gonna have some off this plate, because it's...

Yeah. Just jumping in here. Uh, that was amazing. You guys, uh, really cooked. Yeah, thanks, Sam. And Sam, you've been cooking, which, uh, I think restores-

Thank you

... some faith in, uh, American, uh, dynamism.

Yeah.

Uh, we, we have three very special guests, including, uh, one person that you know particularly well. Come on in, guys. Amazing. There's a- Okay ... middle here. And I guess Sam is the one who- Wow ... kind of tinkered with this.

You can also eat as a- Oh, okay, okay ... cog. But yeah, this is dev unit, so he kind of has, like, a Claude code for a stove. And yeah- Wow ... he's been working on it, so he's the mastermind behind everything.

Great. Yeah. Good job, guys. Okay. Scallops. And you guys can all just take some in.

It told me the temperature to cook for.

What?

So if you get food poisoning, it's Claude's fault.

Okay. Nice. So you guys cooked it on the Impulse stove? Yes. Wow. So it's an induction stove. Wow, wow. It smells really good. This is very different.

78 out of 10, dude. It's amazing.

I feel like... Did you guys use butter on that? Yeah, we used a good amount of butter.

Nope. Shit.

Do you see that stick? That was a full stick at the beginning.

Oh, okay. We used cal- half a stick of butter. Yeah, yeah. Nice, nice.

I love it.

I'm gonna try this. I love it.

I love it. The French way. I'd eat this all day.

Okay. Great. Claude did this. Amazing. Yeah, so an approval-

Good job, Claude

... of the Impulse stove. Yeah. Yeah, Claude. Great.

Claude and Impulse together? Man.

No. Yeah.

I do have some xAI credit, so we'll, we'll try Grok next time.

Okay. Yeah, maybe a voice feature as well to talk back to you while you're cooking. Very true. That would be great.

Oh, I'll get a different voice. Definitely like unhinged mode. Unhinged mode. Yeah.

Yeah, unhinged mode.

I, I gotta do it. I gotta do it.

Great. Well, thank you again, Sam. This is amazing. Yummy. Like, amazing place, amazing demoing.

Thanks for coming through.

Yeah. But best of luck-

And-

... with the launch and everything.

Yep.

And everyone can find it where? Impulse Labs.

You can buy... If you buy direct from consumer, we're shipping now. People have had them for, like, a year. ImpulseLabs.com. But, uh, we're also launching nationwide in Zephyr, in showrooms that have Zephyr products. Um, and we'll be in a couple hundred showrooms- Wow ...

basically by the time this video drops.

Great.

So you might be able to see us in your, in your neighborhood.

Yes. Woo! Go check those out. Awesome. Thank you. Awesome. Yeah.