Arvind Veluvali, Fresco: CEO @ Fresco (YC F24) | AI for Construction

Welcome to another Founder Wisdom Bot YC edition. We have Arvin Velvali. Velvali, I always have problems pronouncing your last name, Arvin, but Arvin is a CEO of Fresco.

Uh he's part of YCF24. Uh for those that know YC, it's the best uh accelerator for startup and incubators. It's located in SF or near SF.

And yeah, Arvin started Fresco. Uh it's an AI for construction company. Uh previously was chief of stack and director of partnership at January AI.

It's a pre precision medicine company based in the Bay Area. Um he also worked at vault health. He was co-founder at down ballot solution in uh 2020 and just an interesting guy altogether and and obviously worked at uh NAS.

I think that was his first job synthetic biologist. So very interesting background. Uh I'm super honored to have Arvin as a guest today, but also as a client of mine.

So Arvin, if you want to give us a quick uh introduction about yourselves and tell us what you're up to nowadays. Yeah, cool. Well, thanks Charles.

Um my name is Arvin. I am the CEO here at Fresco AI. Uh basically we help construction companies, specifically general contractors, put together their on-site documentation more easily.

Uh, we do that using generative AI and we've taken a lot of workflows that have been really annoying because they've traditionally only existed on the desktop and move them over to the phone leveraging the latest in technology to finally make that possible. We're working with some of the largest GCs in the country right now, but um, you know, signing more clients every day, right? And tell us about how you started at NASA in 2020 and how you went from there to a YC company.

Is that a common path? Yeah. Um well so you know originally in college I was actually premed and so I had done a few internships at NASA and um you know those were basically in service of my premed ambitions.

Um I was a biologist there. Um basically uh doing a lot of kind of genetic editing stuff um on uh fungi mostly. Um and the idea was uh I would take that knowledge um from NASA all that experience at the benchtop and um use it to become kind of a a physician researcher.

And so I graduated in 2020 um and I decided to instead of metriculating directly to medical school take a gap year um where I would you know work at NASA but also um I would do my masters in biotechnology at Brown. Um during that time um because it was a pandemic I had a lot of free time and so I ended up starting um Down Solutions like you mentioned with a few of my friends from high school actually and that company I won't talk specifically about what it did but it ended up doing pretty well and being pretty successful. Um and the product itself wasn't necessarily something that was super super complicated to build and I realized that um you know entrepreneurship was something that I was really interested in probably more than medicine.

it was something that really appealed to me. And so I ended up um you know foregoing medical school and uh coming to San Francisco where I you worked in venture capital and then kind of sought more operational roles. So I worked as a product manager and then for the last two years a really great learning experience um as a chief of staff at this uh leading health tech AI company called January AI.

Um and learned a lot and the journey to Fresco um I think you know a lot of people tend to hop on trends in the technology industry right where you know crypto was a trend and suddenly everyone was an expert on crypto and uh I guess that was like the last big one that really defined the the zero interest rate period. Um, I never really saw the use of crypto. It seemed very hypy to me.

Um, in 2022 though, AI was becoming a thing. I remember in November of 2022, Chachbt launched it. Immediately, my mind was blown.

I started using it for everything. And there were people who I followed who were these tech luminaries who said unironically that AI represented a complete sea change in technology. It was the most significant technology advance since probably the internet.

And that really stuck out to me and I thought to myself, look, if I had founding ambitions, like this is a technology to do it with. Like if I had been around for the dawning of the internet, I probably would have wanted to start an internet company. And here I am at the dawning of the AI, I guess, renaissance.

I should be starting an AI company. And so, um, I met my co-founder at Kil and we basically started looking for, uh, different sectors of the economy where AI could really make an impact. and construction.

Um it became evident was for a number of reasons was a really strong candidate for uh that sort of solution. So we've been building fresco for about the last year or so. Yeah.

And we'll get to that. I want to double click on your crypto comments. So at first I didn't get crypto as well.

I think still 90% plus of actors in the field are uh high type of people, right? They're not like value first. Crypto to me is just freedom about banking freedom, right?

like when you start understanding that your bank is not giving you optimal conditions, you know, and when a transfer takes four days instead of uh 20 seconds, right? You you sort of like, okay, we we need a better solution. I also think um and you've been in politics, right?

You've probably seen that. I think uh the banking sector is very much a gerontocracy. Uh there's a lot of old people at the top that control the full system and that's quite frustrating and I think like newer generations want something new.

So, uh I still have uh probably 13% of my clients are in crypto, but like they're very serious and they're total nerds. Uh they're buying they're not like um you know the NFT hypers and I think we've seen a couple of um startups at YC in the crypto space, but I don't think it uh lasted super long. I mean, but yeah, AI totally faking gamechanging.

Uh why do you think you got hooked so quick to chat GPT? Is it because you're uh constantly about self-improvement and you wanted a brainsparing partner? Yeah, I mean I think chat GBT felt like a totally new interface with which to interact with technology.

Um I think you know I was constantly on Google like looking up answers to very specific questions that I had. And with Google it was basically I mean it is an aggregator where you ask it a question and Google doesn't necessarily give you the answer or at least it didn't back then. um it would give you a list of links that might contain the answer.

Whereas with chatbt you biased links of course. Exactly. Yeah.

Well, you could you could ask chatbt a question where um you know if if I had an employee and I asked a very specific question, I wouldn't want the employee to say well you can check these 10 sources. I would want that person to give me the answer, right? And that's exactly how I interacted with chat GBT.

And it made all of my queries like being a curious person like so much more efficient. I was able to more effectively get the information that I was seeking. And so, um, seeing that kind of initial kernel that strategy started as and then finally seeing it kind of morph into this more all-encompassing, uh, tool that has manifold applications, it's been really exciting to see, right?

And I think there's multiple phase in getting real deep in the AI revolution. That number one phase is being a hyper user of Chad GPT and not all of these LLMs. I think Chad GPT is the most stable one.

that altman deploys like crazy always ahead of the curve and also like for example his last update about taking all your memories right that that's huge like all that I ever said to it is now relating to that that's just game changer I think the second step not sure about you but for me it's been it's been building up and we're seeing vibe coding right now that's taking off like crazy to me even before vibe coding I think is just creating pure chrome extensions and connecting uh chat GP PT to improve pretty much everything that I'm doing on a daily basis. Have you started doing that on your end as well? Uh I mean I think chat GBT itself has so many different like uh I forget what they're called like they're called like I think they're just called GBTs where like people make custom GPTs.

I made a few of those of my own. Um I think I kind of maybe skipped the Chrome extension part. Um I think those have proven to be so useful that um that's kind of where I focused a lot of my own kind of noodling.

Um, but yeah, I totally agree. I think I think um I think one of the cool things about AI is it's like really democratized the ability to interact with these technical interfaces in a way that wasn't possible before. And with VOD coding, do you feel that it's going to be possible, for example, for a YC company to go fully solo and be full stack builder and seller?

Yeah, I mean, I think um I think yes and no. Um, I think that with vibe coding, it's really good at getting you to kind of like a technical MVP. Um, I do think there are a lot of things that um, maybe vibe coding can't necessarily account for.

Um, there was this thing on Twitter a couple months ago where someone had vibe coded their entire application. Yeah. What's his name again?

That guy. Yeah, but they got hacked like immediately and weren't totally sure what to do. Um, I think uh I think you know my co-founder is very technically gifted.

I don't think we'd be where we are if like he had vibe coded the entirety of fresco. Right. I'll get his name.

I have it in the WhatsApp convo of mine. Poor guy. Yeah.

I think um that's been a main insight of like coding stuff is that it it needs to be secure, right? Like uh even like building an app that's connected with open API, open AI's API, you don't want to leave your API key in there and like bunch of details like that. And that got me to understood cyber security as a an industry a bit more.

But why did you specifically choose construction as an industry? Did you had experience in there? Do you need to have experience or can you just beat anyone and jump with two feet and be successful?

Yeah, I mean uh I didn't necessarily have any experience in construction. Um I think that to kind of answer your first question like I think you can kind of get into any field as as as long as you kind of meet the right people who do have that experience. I think people are very generous with their time and their insights as long as you approach it with with some degree of like humility and a learner's mindset.

I think if you prove to people that you're there earnestly with uh you know a real ambition to learn and uh gain from their insights, they're a lot more willing to talk to you than you might think. Um in terms of why we chose construction, it was a couple of things. Um, I think the main part of it though is construction represents just uh an outsized uh part of the economy that really hasn't been touched by tech in a substantial way.

And even though there's been a ton of advances in technology, you construction is 8% of the economy. We create like $2 trillion worth of structures every year. And if you look at these uh legacy solutions, a lot of them look very antiquated.

And you know the the bestin-class is a solution like Procore which definitely has a lot of capabilities but even they uh haven't penetrated the entire market in a way that um you know tools like uh Microsoft Office or ubiquitous in fields like uh pharmaceuticals or um you know sales and so we thought there was a really big opportunity for tech to penetrate construction in a robust fashion and we thought AI was the interface to do it because the problem with tech penetration and construction is a lot of the workflow flows um that you know people are trying to transpose to construction are modeled off of workflows that uh exist solely on desktop, right? If you're creating reporting, if you're doing bidding, like you're doing that sitting at your laptop or your desktop computer at a desk somewhere, right? Construction is different because there is seldom time when you're at your desktop.

You're doing a lot of data capture, but it's just happening on your phone. And AI for the first time makes insights that were only um attributable to the phone, only gatherable on the phone. We can finally take those and conduct desktop level work using the phone as an interface.

And we thought that was a really exciting application of this technology. I see two modes in Freso and you tell me if I'm wrong. U first one is that people that may start companies in that space, your competitors, they're like ex construction people now founders.

And the second one uh and by the way the advantage here is that yeah you're like a you're super nerd right you're you're young you bring like fresh new ideas in that and this the second one is AI age related I believe one of the last sectors to be physically disrupted by uh AGI would be construction what do you think about these two modes yeah I mean I think yes um I do think like having innovative ideas is is probably uh emotes like AI just to ship faster. Yeah, you can iterate a lot faster. Um I think honestly the the biggest moat that we have is just the level and this goes back to the the conversation that we were having before about um you know being technical.

The biggest moat that we have is the relationship that we have with our customers, right? Like with AI um you can very easily spin up uh you know mockups, prototypes, whatever of of new technology, right? Someone I watched someone code a clone of Airbnb in like three hours.

Right. Um the real functional one. That's right.

Yeah. Not the one. Exactly.

Yeah. But again, the real mo is is do you have trust with your customers? Do you have those human relationships you're able to build?

And that's something that we index really heavily on is making sure our customers are getting value of our product. Making sure they uh they like us and they like Fresco's technology. And um uh there was this uh this quote I really like which is that your talent opens the door but your character holds it open.

I think that um being really high integrity with our customers and and proving that we're there to serve them has really uh been good for us and been good for them as well. And is isn't it hard for you as a techie kid to go physically and meet them on a construction field? I mean, yeah, it's it's it's hard and honestly like at times like it's it's not very fun, but um you know, I think the other moat is you you do have to probably sometimes get up at 5 in the morning and go to a construction site in like rural Indiana.

Um which I I've done and we've done at Fresco. Um a lot of people don't want to do that, right? You know, if you're selling SAS, you can sell SAS behind your desktop.

You can't really do that in construction. And um I think that's been a big part of the uphill climb, but I do think that's something that because a lot of people aren't willing to do it. Um a that drives trust with our customers and b I think that just keeps out a lot of people who maybe aren't as serious about this.

Yeah. But nothing compared uh to that Airbnb founders at first that had to staple their logo to cereal boxes and and knock on doors and be like, "Hey, why not have someone on an a stranger in an air b air mattress in your uh New York apartment, right?" Yeah, it's pretty hardcore. Um we we definitely did similar things uh when we were kind of initially starting Fresco.

We did like walk onto and get kicked off of a number of construction sites. Um but uh yeah we uh we we definitely did a lot of that that door to door and we we we still do that actually. And where does that resilience comes from in you?

Is it just like pure YC inspiration or did you had that in you from from a young age? Um I think it's a little bit of both. Like I think that um one of the things I was really cognizant of was I wanted to train myself to be comfortable with um with just getting rejected.

There's this one guy uh this is a very popular like internet post I think a decade ago where this person wanted to become really comfortable with rejection. So he would just ask a bunch of audacious things to people over the course of like a 100 days and get rejected at subsequently bigger and bigger and bigger things. Um but what he found was that like if just by being willing to ask for things uh and know that he could likely get rejected he was actually able to like talk himself into some pretty amazing situations.

Um, and so I think that um, one of the things I've tried to do is just be in a position where I'm getting rejected more than feels comfortable because I think if you're getting rejected, you're stretching yourself. And I think when you're stretching yourself, you can achieve those those asymmetrical outcomes. I think that guy was Noah Kagan.

Uh, there's various that did that experiment, but Noah Kagan, yeah, he also did like all these videos knocking on rich people's doors and asking them what they do for a living. He also got invited to yachts and to private jets using the the same method. And just for the sake of doing the experiment, it might be worth it because Noah is already a a deca millionaire for sure.

Um question for you. Do you do cold calls? It's quite surprised me to the other day.

Is that part of the same experiment? Uh it's not necessarily part of the same experiment. Um I think cold calling is one of those things that like uh well in construction like people walk around sites with their phones which is uh one of the common ways that that um we get in touch with with early customers.

I mean the other thing is like uh getting sending a cold email often to the void and like getting rejected. Um I guess it probably hurts less than like getting cursed out on a phone call which can and does happen when you're making cold calls. Um guys, that's right.

Yeah. And so I think that like uh again like it there is a high chance of rejection and there is kind of a a larger uh I guess hurt factor when that rejection comes because it's like a person yelling at you instead of like no email response or someone responding to you via email you can just delete. Um but I do think like there is asymmetric upside to cold calling.

Like we've gotten a a decent number of clients from from cold calling and uh it's definitely something we still do here at Fresco. Right. and tell us how hard it's been one to maybe not to get through YC I think that's been the easiest part but just getting through YC demo day uh getting this off the ground how hard has it been and could you find happiness in these last couple of months?

Yeah, I mean I think um it's been like decently difficult. Um I think it's mostly just because um there's not like a clear path. It's mostly like you kind of have to figure it out on your own.

Um, but I will say like hard and fun are kind of the same thing. Um, I think I've worked in a lot of jobs where they were hard. They weren't necessarily fun.

Um, I think like this is probably fun because it's hard. Um, and you get kind of get to be the the driver of of your own car, if you will. And um, I don't know.

I think this is like the most fun I've ever had at a job. Um, like it doesn't feel like a job, even though I guess technically it is. Um, yeah, it's it's really genuinely a great time.

And is it because your last jobs have been hard at working at the VC firm or being a chief of staff? No, I mean I think like there were definitely moments when I had fun in those jobs and and like you know for example my chief of staff job I learned a lot like um I had a real a lot of really great experiences. Um I think the reason I think this job is so fun is because um all of the experiences that I've kind of cultivated in my early career have been in service of wanting to start my own company one day.

And um I think sometimes just taking a step back and realizing like what a privileged position that that I and and we are in at Fresco um it's it's hard to believe sometimes. And I think that uh the knowledge that I get to wake up every day and you know work with one of my best friends and uh work on problems that I really care about with customers that I genuinely enjoy. Um I mean who who gets to do that, right?

It's it's really great. What does that dayto-day in your life looks like? Um I think it can it can vary.

um which is one of the the nice parts about working in a startup. Um I spent a lot of time on customer acquisition, customer support, uh putting together collateral for various people. Um uh yeah, it it really depends.

Wake up at 5. No, no, I don't I definitely don't wake up at 5. I usually am up around like 7.

Um and then I probably go to bed at like midnight. And you're a musician as well. Where does that fit in your day?

My what? You're a musician. I am musician.

Where does that fit in your day? Do you jam every day? I usually try to actually.

Yeah. Um my uh my last boss when I was chief of staff actually gifted me uh it's like one of the nicest things that he's ever done for me, but she gifted me this very very nice uh like piano keyboard. Um I've been playing the piano for like more than 20 years now and it's like one of my favorite things to do.

Uh it's a great way to like decompress and like wind down at the end of the day. And so um it's not like a conscious thing that I fit into my day, but um you if I got like spare moments and just want to unwind, like it's it's one of my favorite things to do. Once you exit fresco, let's say in three to five to six years, do you want to get back into medicine again?

Maybe. I mean, I guess never say never. Um I mean, I think biotech is is a really interesting field.

There's a lot of cool things that you can do in biotech. you can impact people's lives in a direct way that's not accessible in a lot of other fields. Um, I do think with biotech, uh, it's really hard to be a startup in biotech.

Um, and the reason I say that is because when you're a startup, you really focus on those like iteration loops or feedback loops, right? Where you launch something, you get a bunch of feedback quickly, and then you kind of go back and iterate. And the tighter those feedback loops are and the more experiments you can run, the more successful you can be.

Um there's a quote I really like which is um if you knew you had to fail 30 times before you were successful, how many or how quickly would you want to fail, right? And so I think like fail fast is like a startup motto. Um in biotech like the timelines to running an experiment are so long where you know in startup land like when you're making a a software company uh you can run an experiment on a product inside a week, right?

In biotech it might take you like a year to get a candidate drug. Then you have to like get into humans or you know animals first and then humans and then various stages. It can be a decade before you even have the answer to your experiment right which is really difficult um to do as as kind of a startup.

It requires tremendous capital. It's typically why you see a lot of new biotech innovation come out of like university labs um that can shoulder the cost of this funding that do more basic science. But um I think for now for startups like biotech is a really difficult proposition to entertain.

But aren't they going to have a Chad GPT like revolution in which uh we're we're going to get insilico and we're going to be able to 100x experiments through I really hope so. Yeah. I mean I'm I'm really excited by advances like uh like Alpha Fold.

Like I think that's that's really tremendous. Um I'm definitely a techno optimist. Like I think like uh when when uh I forget who was this uh the CEO of Deep Mind was on 60 Minutes um last week and he was saying in 10 years we could have a cure to all disease.

And I definitely think that's feasible. Like if you look at cancer treatments, most cancer treatments that we have today, uh I'm 26, those were not available when I was a kindergartener, right? And we're getting more and more and more every day.

Um and uh you know, we we've cured cystic fibrosis, we've cured uh um cickle cell, we've cured hered hereditary deafness with tools like crisper. Um it's a combination of a couple things like one, how quickly can we adopt this new technology? And two, how quickly can the regulatory landscape catch up to what's happening in science.

You know, when I mentioned we've cured single cell and like CF like these are physically cured with tools like crisper. We're waiting on the regulators to catch up and say yes, this is now a treatment that everyone can have access to. And I'm bullish on Dennis and Sam as well, but don't you think that them being absorbed by big corporations is drastically slowing them down by Google and by Microsoft?

Uh, I don't know. like I I I'm don't really know what goes on inside Google or DeepMind. I do think like uh you know when I talked about like universities doing a lot of the foundational research that leads to innovations.

I mean I think Google's kind of the same right Google invented the transformer that that led to like GPT. Um and so I think like these large organizations with a lot of capital can certainly drive a lot of innovation. I I do think they've had more of an innovative posturing over the last uh year or so if not longer.

Um so I'm optimistic like I think the pie is is plenty large and can get larger still. And I'm also a techno optimist. Do you feel that we're going to get an AGI that's going to bring massive value to the world in two to three years but slash 80% of the John market?

Yeah, I mean I think like the the question around like AGI, I think the goalposts are continually shifting. Like what is AGI? Like how do we define it?

Like I don't know. But I certainly think like we have AI right now that is probably commensurate with the capabilities of like the average person, right? Like Chat GBT can probably answer a question just as well as like your average research analyst.

Uh in terms of like cutting the job market, like I I don't know. Um, I really can't can't comment. Like I know a lot of people have very fatalistic outlooks on the job market uh in the next like 20 years.

Like will we even have jobs? Uh I do think like we'll probably have different kinds of jobs that show up. Uh we might have some sort of contraction but um I'm confident like I'm optimistic that uh you know we we might have a solution like a UBI or more time for leisure or um you know something else but it's hard to comment.

Yeah. two pieces of definition just to um see what you have to about these on AGI. One, it's agentic and we're about to hit that revolution, right?

So, it has autonomy over its action right now and all of these tools don't really have uh agentic possibilities. And second, it would drastically outperform humans. Does that mean it would replace jobs?

Let's say that I'm an ICO and and I still want to give some work to fellow humans. Sure. But I don't know if I'm really really greedy.

I think I would just have like agentic solutions all over the place. So what do you think about these? It's a good question.

Um I think a couple of things like one accountability um is is really difficult to force upon a robot um if you don't have like a human uh kind of undergirling that. The other thing is um I I think like before we used to have like secretaries typing everything out physically and secretary was a job and then everyone got a laptop or they figured out how to use their own typewriters and they just became more efficient like typing wasn't something that consumed an entire day of work for a person but uh the economic output we've been able to achieve just been drastically higher. Um, I think a AI is going to be a very similar thing where you can have this like team of minions doing work for you that um is is otherwise difficult or ownorous for you to do where they can communicate with your suppliers.

They can, you know, pay your rent, right? Or they can uh maybe even negotiate with other AIs. Uh the other thing I'll say with with agentic AI is I don't necessarily think AI is going to be maximally useful without some sort of physical embodiment, right?

Like you know, look at medicine for example. Like I think that like I would honestly probably trust an AI's diagnostic capabilities given the right tests more than like the best doctor because AI has more information than any human can ever absorb. But that being said, like there is still a human who's drawing my blood, there's a human spinning the centrifuge and doing all the tests.

Uh you know, if I have to have a procedure, it's a human doing that. like AI can be valuable in in kind of assisting in those workflows but in so far as there's like an interaction with the physical world like it's really difficult to imagine AI being uh replacing any of those workflows right yeah and these professions would become more psychotherapists like so for the experience and also the trust factor the human connection oh there's a human it's kind of silly right it would be like a full AI backing but humans trust humans yet they don't trust fully AGI but I I do think uh that there would be moment in which they would trust more an AI than their their human counterparts. Uh if statistics are right, I also think an AGI would be unboxed and uh fully unhinged, not aligned with humans, but then would it even want to serve us?

What do you think about that factor? Uh yeah, I mean I that's a really interesting question. Um yeah, I I I don't know.

I I think constraints around AI are a really interesting thing to contemplate. Um, I I would caution against letting AI kind of take over a lot of the capabilities that we as humans execute. Um, I I remember I had a math teacher in in sixth grade.

We were uh taking algebra and we got access to this like top-of-the-line graphing calculator and she said, "Here's a calculator. It's really smart. The thing you got to remember is you can never let it be smarter than you." Um, and I think AI is the exact same thing where, you know, it's plenty smart.

It can have a lot of capabilities, but if we're using its capabilities as a substitute for our human judgment, then we're in trouble. So, how we keep AI like subservient that I'm not sure. I just think that's maybe the guiding principle that I'd follow.

One last barrage of rapid fire questions because I know you need to run. Um, when are we going to get AI into robots? Is Fresco going to take advantage of that?

And would there be a $1,000 home in 5 to 10 years from now? Uh, when are we going to get AI into robots? Um, I would hope within the next decade.

Is Fres going to be part of that? Like, yes, absolutely. Um, will we get a $100,000 home?

Like, yes. I think that's already possible to do $1,000 or $1,000 home. Like, yes.

I think that's already possible to do through materials like, you know, uh, uh, 3D printing or even uh, you know, soil compression. The only question is like can we get the regul again again can we get the regulators to move as quickly as the techn is moving like here in California the cost of building a home is not the cost of materials necessarily it's the cost of dealing with regulators and until we can get around that it's really difficult to see a way we can get the u the cost down that much all right Arvin thank you so much where can we find out more about you great yeah you can find more about me uh at the fresco website it's fresco-ai.com so that was Arvin Velu Valley and Charles founder founder