How Non-Coders Are Vibe Coding $100K+ Businesses with AI
Replit's head of product engineering walks through three non-coder founders who built real six-figure businesses, then live-demos the four steps that turn a raw prototype into one.
Posted
1 weeks ago
Duration
Format
Interview
educational
Views
8.5K
69 likes
57 · 43
Big Idea
The argument in one line.
As AI drives the cost of writing code toward zero, the businesses that survive are the ones built on unique expertise, trust, security, and distribution, not the code itself.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
You've built or want to build an app with an AI tool like Replit, Cursor, or Claude Code and don't know what happens after the first working version.
You have deep expertise in a niche (fitness, peptides, interviewing, local business marketing) and want to turn that knowledge into software.
You run a company and want a concrete case for replacing a SaaS subscription with an internal tool built in-house.
You're deciding whether to build your product on top of one AI model or stay model-agnostic.
SKIP IF…
You're already comfortable publishing, securing, and monetizing production apps, the four-step process here is introductory.
You want a step-by-step coding tutorial; this is a strategy conversation with a screen-share demo, not a build-along.
TL;DR
The full version, fast.
Amol Jain, Replit's head of product engineering, tells Peter Yang about three non-coders who built real businesses with AI: a college student's peptide-tracking app that made $60K in month one, a repeat founder's AI-certification platform that hit $180K in two months, and a pro gamer's local-creator marketplace now doing six-figure ARR. His thesis: as the cost of writing code falls toward zero, value shifts to unique expertise, security, distribution, and trust. Amol then live-demos converting a raw AI interview-coach prototype into a real product, buying a domain, running a security scan, adding Stripe subscriptions, and running an SEO scan, before discussing why companies are replacing SaaS with internal tools and why staying model-agnostic now matters more than picking the single best model.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Cold open teases a founder's story (an agency quoted him $100K, he built it himself in three days and hit $180K+ in revenue in two months) before Peter introduces Amol as Replit's head of product engineering.
02:08 – 08:04
02 · Three vibe-coded apps that became real businesses
Amol walks through three non-coder founders: Cedric's Pep AI peptide/GLP-1 tracking app ($60K in month one), a repeat founder's AI-certification platform ($180K+ in two months), and FaZe Apex Yusuf's Try Nearby, a local-creator marketplace now in YC's Summer 2026 batch with $100K+ ARR.
08:04 – 12:31
03 · What six-figure vibe-coded apps have in common
Amol argues the winning apps aren't generic tools like meal planners, they lean on the founder's unique expertise, judgment, community, or network; a Riverside sponsor read runs mid-chapter.
12:31 – 13:59
04 · Live demo: turning an AI interview coach into a real business
Amol opens a raw Replit-built landing page for an AI product-management interview coach and frames the session: turning this 'raw product' into something production-grade.
13:59 – 17:50
05 · Steps 1 and 2: publish the app and make it secure
Amol buys the domain coachpeter.net, publishes the app, and runs Replit's automated Security Center, which builds a threat model and scans for exposed secrets, malicious packages, and PII leaks.
17:50 – 21:35
06 · Steps 3 and 4: add payments and optimize SEO
Amol asks Replit's agent to add $100/month and $1,000/year Stripe subscriptions with no manual Stripe setup, then runs an SEO agent to make the app discoverable by both search engines and AI answer engines.
21:35 – 28:22
07 · How to replace expensive SaaS with internal tools
Amol gives two internal Replit examples: a sales-demo tool built by a non-coder on the RevOps team in two days, and an in-house BI analytics platform built by a data scientist to replace a costly analytics vendor.
28:22 – 31:02
08 · What software survives when code costs almost nothing
Amol's test for durable SaaS: if the cost of code goes to zero, what's left to pay for is trust, data, infrastructure, atoms, labor, and network, citing Salesforce, Workday, and DoorDash as examples.
31:02 – 36:06
09 · Why value is moving from models to apps
The conversation closes on hot takes: frontier models are commoditizing, so app-layer differentiation is the 'last mile' (hosting, security, distribution), and staying model-agnostic protects against pricing and policy shifts from any single AI lab.
Atomic Insights
Lines worth screenshotting.
A 22-year-old with no coding experience built a peptide and GLP-1 tracking app that made $60,000 in its first month and now has tens of thousands of active users.
A repeat founder was quoted $100,000+ by a dev agency, built the same AI-certification platform himself on Replit in three days, and hit $180,000+ in revenue within two months.
A professional gamer with no technical background built a marketplace matching local creators with local businesses; the company is now in Y Combinator's Summer 2026 batch with $100,000+ ARR.
The vibe-coded apps that succeed aren't generic tools like meal planners, they're built on a founder's specific expertise, community, or network.
The test for what software survives as code gets cheap: if the cost of code goes to zero, what's left to pay for is trust, data, infrastructure, physical assets, labor, and network.
Vibe-coded apps are notoriously exposed to attackers, so an automated pre-publish security scan that builds a threat model and tests for it has replaced a two-week, paid security audit for many builders.
Publishing a vibe-coded app used to mean sharing a localhost link with friends; hosting platforms have made taking a prototype live as simple as buying a domain and hitting publish.
Adding Stripe subscription billing to an AI-built app can be as simple as asking the agent to add a $100/month or $1,000/year plan, no manual Stripe signup or KYC required.
An SEO scan run on an AI-built app also optimizes for AI answer engines like ChatGPT, not just traditional search, treating discoverability as its own growth step separate from the build.
Replit's own sales team got a demo tool built by a non-coder in two days that saves what comparable commercial demo software costs, often six figures a year.
A Replit employee who was a data scientist but not a coder built an in-house BI analytics platform on the company's own BigQuery data, replacing a SaaS analytics tool that would have cost millions.
The 1x employee does their job well, the 2x employee builds a personal tool to do their job better, and the 10x employee builds a tool the whole team can use.
Software categories most likely to survive commoditized code are the ones selling trust, regulated risk-taking, proprietary data, physical infrastructure, human labor, or network effects, not just an interface.
As frontier AI models become commoditized, an app-layer company's real differentiation moves to the 'last mile', reliable hosting, security, and distribution, not the code generation itself.
Staying model-agnostic is becoming a competitive advantage in itself, since it lets a product route to whichever model is cheapest or best for a given task instead of being locked to one vendor's pricing and roadmap.
Takeaway
Four steps turn a vibe-coded app into a real business.
WHAT TO LEARN
As AI makes writing code nearly free, what separates a $60K app from an abandoned prototype is publishing, securing, monetizing, and optimizing it, on top of a founder's own expertise, trust, or distribution.
02Three vibe-coded apps that became real businesses
Domain expertise beats generic ideas: a peptide-tracking app, an AI-certification test, and a local-creator marketplace all came from founders who already lived inside that niche.
A repeat founder who was quoted $100,000+ by a dev agency built the same platform himself in three days and reached $180,000+ in revenue within two months.
None of the three founders had coding experience before using an AI agent, and all three grew past a hobby project into a business with real, recurring revenue.
03What six-figure vibe-coded apps have in common
Generic, commoditized ideas like meal planners and exercise trackers are getting swallowed by everyone building their own version with AI, so a durable app needs a founder's specific edge.
The overall market for new businesses is still growing even as software gets easier to build, because more people with niche expertise can now turn it into a product.
Vibe coding still requires marketing and distribution; building the app is necessary but not sufficient, reaching the audience that needs it is the other half of the work.
05Steps 1 and 2: publish the app and make it secure
Sharing a localhost link with friends is treated as an industry-wide joke; production readiness starts with buying a real domain and using a hosting platform that abstracts away servers, databases, and auth.
A pre-publish security scan builds a threat model for the specific app, then tests scenarios like exposed secrets, malicious dependency packages, and mishandled personal data before anything goes live.
Security scanning also needs to run on an ongoing basis as new features get added, not just once before the first launch.
06Steps 3 and 4: add payments and optimize SEO
Adding subscription billing can be as simple as asking the AI agent for a monthly or yearly plan; the platform handles Stripe account creation, KYC, and sandbox testing behind the scenes.
An SEO scan optimizes discoverability for both traditional search engines and AI answer engines like ChatGPT, treating growth as a distinct, ownable step rather than an afterthought.
07How to replace expensive SaaS with internal tools
A non-technical RevOps employee built a sales-demo tool in two days that replaced software costing six figures a year, because bespoke internal software isn't beholden to another company's roadmap.
A data scientist with no coding background built an in-house BI analytics platform on the company's own BigQuery data, avoiding a costly analytics vendor that didn't fully fit the use case.
The progression from 1x to 10x contributor is building a personal tool, then building a tool the whole team or company can use.
08What software survives when code costs almost nothing
The test for durable software: once the cost of writing code approaches zero, what's left to pay for is trust, proprietary data, infrastructure, physical assets, human labor, and network effects.
Salesforce survives as the system of record for sales data, Workday survives because it carries regulated payroll risk, and DoorDash survives because it moves physical food, not because of their interfaces.
Even SaaS incumbents are adapting by going headless, exposing APIs, CLIs, and MCPs so AI agents can use their systems without anyone touching the actual UI.
09Why value is moving from models to apps
As frontier AI models become commoditized, an app-layer company's differentiation shifts to the 'last mile': reliable hosting, security, and distribution, not the raw code generation.
Staying model-agnostic protects a product from any single AI lab's pricing changes, policy shifts, or outages, and lets different tasks route to whichever model is cheapest or best.
Enterprises have stopped chasing the highest-benchmark model for its own sake and started asking what return a given model's cost actually delivers.
Glossary
Terms worth knowing.
Vibe coding
Building a working app by describing what you want in natural language to an AI coding agent, without writing or deeply understanding the underlying code yourself.
Replit Agent
Replit's AI coding assistant that writes, runs, and iterates on code from a user's plain-language instructions inside the Replit platform.
Security Center
An automated Replit feature that builds a threat model for an app, tests it against likely attack scenarios, and reports vulnerabilities before the app goes live.
SEO agent
An automated Replit tool that audits and improves an app's discoverability by both traditional search engines and AI answer engines like ChatGPT.
Answer engines
AI systems such as ChatGPT that return direct answers to a question instead of a list of links, making them a new discovery channel alongside search engines.
Model-agnostic
A product or platform design that lets a system switch between different underlying AI models rather than being built around a single provider's model.
ARR
Annual recurring revenue, a standard metric for the yearly value of a subscription business's recurring contracts.
“Agent could pick up a package that... there's so many malicious packages out there. It could install a malicious package that steals stuff... agent, for example, could be leaking PII because it's not stored in the right way.”
concrete list of what a pre-publish security scan actually catches→ newsletter pull-quote↗ Tweet quote
21:45
“The whole sort of past 10 years led to a lot of SaaS startups that were just essentially interface over some data.”
sharp, quotable indictment of a decade of SaaS→ newsletter pull-quote↗ Tweet quote
28:00
“The 1x person is just thinking about doing their core job... a 10x person is building a tool that the entire team or the entire company can use.”
clean three-tier framework stated in one breath→ newsletter pull-quote↗ Tweet quote
29:47
“A DoorDash is valuable. It is, you know, actual humans are bringing you your food.”
concrete, funny example of a business that survives the zero-cost-of-code test→ IG reel cold open↗ Tweet quote
33:24
“It's almost great to have that indirection layer, that app layer that is model agnostic and that helps you orchestrate and run all your AI workflows.”
names the strategic bet on staying model-agnostic→ newsletter pull-quote↗ Tweet quote
34:45
“It's a funny meme if you've seen of lighting a cigarette with a flamethrower.”
self-contained joke about wasting expensive frontier-model tokens on trivial tasks→ TikTok hook↗ Tweet quote
Topic Map
Where the conversation goes.
00:00 – 02:08sparseCold open + guest intro
02:08 – 08:04denseThree non-coder success stories
08:04 – 12:31Common traits of winning vibe-coded apps + Riverside ad
12:31 – 13:59Live demo setup
13:59 – 17:50densePublish + security scan demo
17:50 – 21:35densePayments + SEO demo
21:35 – 28:22denseInternal tools replacing SaaS
28:22 – 31:02denseWhat software survives
31:02 – 36:06Model-agnostic hot takes + close
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
17px
metaphoranalogy
He went to an agency that quoted him $100 ,000 plus. So he jumped on and essentially within three days was able to build the whole thing end to end. Within the first two months, he was already at $180 ,000 plus in revenue.
If the cost of code goes to zero, what do you pay for? It's this unique expertise that they have. It's the unique judgment or taste.
It's the security. It's the distribution. It's all of that extra packaging that meets the user where they are and users pay for that.
Hey, everyone. Today, I'm really excited to welcome Mo. As Replit's head of engineering, he has a firsthand look at what people have built with AI that actually make money and become real businesses.
And we're also going to talk about how AI is impacting SaaS and how Replit and other app player companies can differentiate from the frontier labs. So welcome, sir. Thank you, Peter.
Great to be here. So now that I'm kind of independent, you know, we were just talking about this. I would love to just...
Make an app that makes money while asleep. And you have a broader view of this than anyone, right? At Replit.
So can you start by sharing some examples of maybe Replit apps to actually make more money? People have actually built real businesses with this stuff. Absolutely.
Yeah. You know, one of the key reasons I joined Replit was exactly that. Replit really lets...
people, you know, it's the entrepreneurial dream of helping tons of people create businesses that make money. And that's exactly what we're seeing is every day we see tons of new businesses start that make real money. And these don't have to be with massive TAM.
We see businesses that, you know, serve a few hundred customers, but these are making real money and helping people, you know, have that upward economic mobility. And so essentially, I have a Few examples that I pulled out.
In some ways, you know, what Droplet does is takes a person's advantage. That advantage might be their distribution, their network, their community, their expertise, their judgment, and turn that into a business. So let's start with one that's called Pep AI.
Cedric. Cedric is someone who lives in Oregon. He goes to University of Oregon.
He's 22 years old. He has never coded in his life before. His roommate and Cedric decided that they'd heard about peptides.
They wanted to take peptides and they got really deep into this community of peptides and GLP -1s. And as he did that, he sort of realized that there was... really an opportunity to help the broader community and turn that into a great business.
He wanted to build something that helps anyone track their peptide and GLP -1 intake and also learn more about these emerging sort of medications and such. And so he jumped on Replit. Again, no exposure to code before.
He was able to build a mobile app. And here's his landing page. Yeah, that looks really good.
And really tapped that need because after a little bit of testing and experimentation with distribution, he was able to grow this product very significantly. He was making $60 ,000 in his first month. And the app has grown.
It now has tens of thousands of active users and growing every day. And so it was really amazing to see someone who just had that entrepreneurial spirit and that curiosity. learn a lot about a certain topic and then build a great business.
And he's making money from subscriptions or in -app purchases? He's making money through subscriptions, yes. Okay.
Yeah, that's super awesome. The website looks great. Yeah.
Yeah. Another example that I'd love to walk you through, this is a... A wholly different persona.
So John, he is actually a repeat founder. He had started something in the past as well. As AI was taking off, he sort of realized that there was an opportunity to help people measure their AI proficiency and even get certified for it, especially enterprises and companies could do that.
And he's non -technical, didn't know how to build it. So he went to an agency that quoted him $100 ,000 plus to build this platform for him.
That didn't sit right with him. He had heard of Replit, so he jumped on and essentially within three days was able to build the whole thing end to end. It's something that has this AI proficiency sort of tool that can measure a person's AI proficiency, way to get certified, et cetera.
And he built and launched it. And within the first two months, he was already at $180 ,000 plus in revenue, purely because as a repeat founder, he had this great product intuition. He knew how to distribute and sell.
He just didn't know how to build. And a tool like Rocket really made it easy for him to do that. Yeah, and now he has a homepage for people to come and check it out and take the test and maybe get a lead and stuff like that, right?
Yeah, exactly. Yeah, 100k, man. Maybe that agency is using Fable tokens.
I don't know why it's 100k. It's very expensive. Yeah.
Okay. Do you have another one? Yeah.
Let's talk one more. Let's see one more. So if you're familiar with gaming at all, if you've ever been a gamer or been in those communities, you definitely know FaZe, the clan that is incredibly popular, one of the most successful gaming groups out there.
And FaZe Apex Yusuf, he's a phenomenal, very famous, and very successful gamer. He had over years built an audience. He's in some ways a creator who streams his video games and such and has had made money through that.
But recently he wanted to start a different business. And again, someone very non -technical, just someone who understands sort of audiences and creators. And what he wanted to build was a way for local businesses.
to connect with local creators to drive that word of mouth for them. It's a topic, you know, he obviously knows a good bit about. And so he came to Replit, built the whole thing here, and he's been incredibly successful.
They're in YC, summer 26 batch now. Him and his co -founder, and for this company called Try Nearby, and they're upwards of 100 ,000 ARR when we last chatted with them. So, so this is like, uh, creators helping local business, like restaurants and stuff, get the word out.
Yeah. It's almost a marketplace that's matching those, those local creators to, to these local businesses and to get that word of mouth out. Yeah.
That's amazing. Yeah. Yeah.
I have a lot of restaurants that I love that I'd love to promote. So check it out. Yeah.
Yeah. This episode is brought to you by Riverside. I've used Riverside for years to record my podcast because it records in 4K resolution each person locally.
So the audio and video still comes through clean, even if a guest's Wi -Fi gets shaky. But the reason I love it now more than ever is what happens after we stop recording. If I go in here, I can use these AI tools to remove pauses, remove filler words, and just clean up the recording.
And I can also edit a transcript directly and it will automatically generate clips with captions ready to publish to YouTube, Spotify, and all types of social media platforms, all from one place. As a one person creator business, that matters a lot. Riverside is the upgrade your content workflow needs.
Try it at creators .riverside .com slash Peter Yang and use code Peter Yang at checkout to get one month completely free. That's creators .riverside .com. Now back to our episode.
Let me kind of stress this a little bit, right? Because like I personally have also built a bunch of apps using Repl .in and other things. Like I built like a fitness app and I built like an app to cure my movie taste and all that kind of stuff.
But I feel like, so like, I guess my question is, do you see any common themes across these examples of apps that actually make money? Like are they targeting a specific very niche problem or is it kind of like, it's not just software, but it's also services? Do you have any tips around some of this stuff to actually build a real business out of this?
Yeah. So I think you're correct that a lot of software, it's becoming rather easy to build software. And so a lot of generic use cases, meal planners or exercise trackers, not to diminish your idea, are getting rather commoditized.
People are building their own versions. Or in some ways, I see... In the future, agents just kind of swallowing all of that whole, everyone has an agent that has a skill that does this.
So I think the thing, that said, the overall pie of businesses is still expanding. If you look at Stripe data, you can see the number of businesses being created and going up. It's because it's that unique sort of...
advantage that that people have and especially the sort of you know average joes in the past that have no exposure to technology it's their it's their sort of like unique expertise that they have it's the unique sort of judgment or taste or their community their their network it's people tapping into those to create the types of things that can be actually tangible businesses.
And these don't have to be massive businesses. Again, I think I really like to emphasize that there is a lot of success you can have with just a small business that thrives. And so I think it's about identifying what that unique advantage is and really leaning into that.
So we see pool cleaners and nurses and jewelers and such, all these people with domain expertise that in the past they wouldn't have been able to turn into businesses now really bring those ideas to life so i think i think the tip is truly finding that that um that specific advantage and i think one thing to always remember users are are people inherently are lazy not everyone has that entrepreneurial spirit and desire not everyone's gonna gonna you know do everything themselves there's people if you offer them value they will pay for it It's kind of like done for you.
Okay, so let me kind of summarize. So I guess there's one aspect is like, what is your unique skillset? What is your unique knowledge?
And also like, you can't just like vibe code this thing and like put it on there and hope to get business. You gotta do a lot of marketing, right? You gotta like reach the unique audience that you have online.
And then maybe just like some additional services or other things that you can offer. Absolutely, yeah. And I think more and more, I see this as...
You know, if you just zoom out and think of businesses in general, the types of businesses that survive are the ones that have sort of either this expertise that we've been talking about or more on the B2B enterprise side. It's, you know, businesses that offer trust or infrastructure or network or systems of record data, even Adams services, as we were talking about.
I think those are incredible businesses that are now accessible to everyone. That makes sense. That makes sense.
Awesome. So let's say I build like a app or like a website on Replit or like another provider. And how do I, like, I can hit a button and take a live, right?
So is it as easy as that? Or should I do some security scans, do some checks before I do it? Like, can you show us how this, what's the difference between a vibe coded app and an app that's actually ready for production?
Absolutely. Great question. And I, what I did was in preparation of this actually.
come up with an example that that we can walk through and and show the audience of of how to go from that raw product that purely vibe coded first version to something that can be production grade and also um also you know be something that grows into a business yeah and so um let me go to this screen so you know you have this amazing podcast where you have product experts come on here.
You have these product experts in your network. And one topic you guys always discuss is product management market has changed. The jobs are different.
The interviews are different. What if, since you were looking for an idea of a business that can make you money, what if you were to build an interview coach where you pair up product experts with applicants so they can... do mock interviewing and, you know, rather than just repeating the circles framework or such, actually give real, you know, I actually can pass these interviews.
So I figured we will do a session here and see how we would take this idea and turn it into a real business. So first of all, let me sort of walk you through how I got to even this page. Essentially, I dumped a very sort of I use Whisperflow a lot.
So I just Whisperflow dictated what I wanted this to be and use droplet design first to sort of explore different directions of what might feel right for a product like this. And I think I tried a bunch of different variations. And I think what I really liked was the simplicity of your landing page.
So I just... took that as the starting point and asked Droplet to build it. And what it did was built this whole landing page with some basic functionality to accomplish what we're looking for.
But this is what I would call that raw product. And now we can sort of focus on how do we turn it into a real thing. Yeah, I think having my face on there can either be trust building or grifting.
I think if you have a trustworthy face, I would purchase. Great, great. Amazing.
So I think first thing is even taking this thing live. You know, you always see that those stories of people use a cloud code and they can, of course, buy code something, but then they're sharing local host links to their friends. That obviously it's almost a meme now because how often it happens.
And I think one great thing about Repl .it is we have the home road infrastructure that. abstracts away all of the complexity of hosting and databases and auth and all of that. So to an end user, our users tend to be non -technical.
It's a rather seamless process. And so let's dive into this. We call it publishing.
And as a part of publishing, I already did publish it, but I'll show you a couple of things. One, I think it's important to have a great domain. We're building a business here.
So we have a way for you to easily... Buy a domain. I bought coachpeter .net.
Coachpeter .com was a little too expensive, so here we are. And then we go on to publishing. But before we publish, one thing that's important is making sure that your site is actually secure.
It couldn't actually survive attackers because there's a lot of bad actors out there that are constantly attacking these. And vibe -coded websites are notoriously... notoriously exposed to those things.
And what we've done is built this, what we call a security center. It's essentially a security scan of your app. In the past, you would have to have a security team that you contract with, pay them a lot of money.
They spend one to two weeks doing this and give you back a report, which you then have to go build it. We essentially built a product that... runs the scan, gives you the outputs, and just ensures that your app is actually secure right before publishing, but then also on a recurring basis as you build more functionality into it.
And so to see how this works, let's say we're just going to hit one button, which opens up a new task and starts running a security scan. What's happening here is the... The agent, Droplet agent, the security agent specifically, is building a threat model for this specific thing.
What are all the ways things could go wrong? And then it's running a deep scan with the different scenarios and testing to make sure that it can detect all of those problems and then surface them to us. And so this is going to take a little bit of time, but I just wanted to kind of show what happens on the security side of things.
Okay. So while this is running, let me ask you this, like, you know, as an engineering leader, what are like the top two or three things that could go wrong? Like, obviously, if I put my API key in the code or something, right, that's like one thing that go wrong.
But is there something else that this thing looks for? There is so much that could go wrong. There's, you could...
Agent could pick up a package that, you know, there's so many malicious packages out there. It could install a malicious package that steals stuff. Secret scanning that you talked about.
There are very clowny things that could often happen as well, where agent, for example, could be leaking PII because it's not stored in the right way. Stuff like that could happen. And so this essentially...
The replica agent that's building it already does a pretty great job of making sure that doesn't happen, but it's important to just tie the knot and make sure that everything is actually secure. And while this is running, maybe we can talk about a few other interesting things that turn this, again, raw product into something that's production grade.
Sure. Yeah, another thing that I would love to highlight, so we'll go back to main while our security scan is working here in parallel and it's scanning. Let's talk about a few other things.
Another thing, you know, we would want to monetize. At the end of the day, a business is something that will actually make money. So you can just ask Repl .Agent to add a way to let my users buy subscriptions.
$100 a month or per year. And it's as simple as that. We essentially have this deep partnership with Stripe where we've abstracted away, again, all of the complexity.
So user doesn't have to go sign up for Stripe, do the KYC upfront, figure out what plans is, what prices are, what subscription is, versus what one time is. All of that is taken care of. We are in the background setting you up a Stripe sandbox, setting up the key, setting up the plans and prices and making the whole thing work so you can test.
And once you feel good, that's when we give you the live Stripe to go and just sub out the last pieces to take it live. And I think we see so many people who've never monetized anything in the past. able to add this functionality now and turn things into real businesses.
Yeah, that's super easy. Do you have to make a Stripe account first or it just kind of works? It makes the Stripe account for you and all you have to do is just go ahead and claim it.
Nice, nice. Another thing I'll touch on, we were just talking about distribution. At the end of the day, so much of the game is distribution if everyone can build.
And so we're investing a lot in making that, in giving you the tools that... Any entrepreneur can try to distribute and get audience for their product. And there's a lot more that's coming very, very soon.
But one thing that I would love to show you is this whole idea of an SEO scan, especially for the tool that we're building here, which is a product management interview coach. That's a very hot term on Google and with Answer Engines. You can run a SEO agent as well that does all the work to make sure that your app is discoverable by both search and answer engines like ChatGPT.
So it really helps you. solve one piece of growth. There's a lot more that's coming in terms of ads and giving people a way to market their product through emails and such.
But this is our first step on this space because we really want to help people not just build great businesses, but also grow those businesses. Yeah, this is a really good, smart investment. I think I'm lucky I built like a big Twitter audience.
And I feel like audience is like the durable asset these days. Like if you have a good audience, as long as you're not grifting, you can like launch multiple products and hopefully one of them will work, you know? So yeah, really important.
Yeah. So, you know, to, to sort of summarize here while, while these agents are running, they'll take a while, but essentially we took this, we took this, this design that we had built, turned it into a real product. And then now we've added.
things like monetization and security and figured out a way to even distribute it. And I think that really encompasses the journey that we see a lot of Repl .it users go through every day. Awesome.
All right. Well, let's talk about another topic. So I think Repl .it has been doing really well in the enterprise market.
And what I'm seeing is that companies, they don't have to pay for all these crazy SaaS products. They can just build a bunch of internal products to replace them. Is this actually happening or is it something I'm making up?
Absolutely happening. We see this every day. The whole sort of past 10 years led to a lot of SaaS startups that were just essentially interface over some data.
And now that the cost of code is going almost to zero, there's real reckoning and people want to build. It's not just about money. It's not just about saving on those costs.
It's very much about building bespoke solutions that fit your workflow as a company. We internally, you know, this is obviously buy as take, but we internally have replaced so many tools with things that we built ourselves that work better and are tailored to our use cases. Do you have some examples to show for that?
Like internally how Replit uses tools? Yeah, so let's see. I think maybe we could talk about...
recent one that I really like. So, you know, as we were just discussing, Replit's now getting really big in the enterprise. On the backs of their success, our sales org has grown a ton.
And what the sales org is doing is demoing to prospects. It's a product that you really see the power of when you demo it to a prospect. But demoing a product like Replit, in general, AI products is kind of hard because A, As we just saw, we started a security scan, but it's going to take 10 minutes for it to complete.
So there's that inherent dead time and lag. And then secondly, they're non -deterministic. Things could go wrong.
It might produce an output that you're not super keen on. Agent might take different directions. And it makes demoing really hard.
So what we wanted was to create a compelling demo that's predictable, that lets someone follow a script and show the right things to a customer. There's tools for this. There's tons of startups that do this.
The products are incredibly average and basic and they're expensive. They can easily run into six digits. And so someone on our RevOps team, again, never written code in their lives, took this idea and over two days built something that now our sales folks are using.
So I would love to show that here. Great. As we were describing, just to kind of share with this product how it actually works, there's a Chrome extension and a salesperson will install the Chrome extension.
Let's say they want to create demo for a specific thing. They'll use the Chrome extension to walk through the product, record screens and annotate things. And then once they're ready, it turns into this artifact that is demoable that you can present to prospects, give it to your salespeople and such.
And so let's look through something. A replica build flow that someone recently built, very core. This is the core interaction that every prospect, we walk every prospect through.
And a couple of other things I'll mention here. Like I was saying, again, it saves on, say, six figures of SaaS spend. But more importantly, it is bespoke software where anytime we want to add a feature.
we just do it we're not at the mercy of someone else's roadmap and so someone asked the creator of this tool hey i would love to have way to prospect to create different demos for different personas so we added a persona feature and so now there's four different personas there's different demos and such i think that is incredibly powerful and just really lets us let's us uh you know solve our own problems uh and not pay someone money for it the other point makes a lot of sense too because as a sas provider if i want to add a feature I have to think about all my other customers too, right?
But if I just have personal software that Repta is using itself, it's so much faster and easier to change. Absolutely. Maybe another example that I really like personally, very recently, as a person who, you know, so much of my job is building products and being deep in the product side of things, analytics is so important.
And so we recently churned off of our analytics. BI analytics provider and built something in -house entirely. And so the data was already ours.
It was already in our BigQuery. It didn't make sense for us to pay a very hefty fee to a product that didn't quite meet our needs in a lot of different ways. And so this is hot off the press as well.
We essentially built something that does exactly the same thing. I'm showing something that's not confidential. It's essentially an admin portal tool.
and how many people are opening it over time. But you can see it gives all the BI capabilities of different types of charts, comparisons, timeframes. You can create funnels or data tables or retention charts and measure different things.
It does all of that, but it's ours. And we've been slowly adding a lot of intelligence to it as well to make it really, really great and solve for our use cases. Yeah, I mean, those analyst platforms cost millions of dollars, right?
It's expensive. They truly do. Yeah.
Wow. Okay, but that wasn't vibe -coded by a single person, right? Because that was pretty complicated.
Is there a team behind that? Believe it or not, it was vibe -coded by one person. In this case, it helps that the person is a data scientist themselves, so they understand it a little bit better.
But they're not a coder. They're not a builder. But they were still able to do this.
That's amazing, dude. And I can see it like... As that product improves, you can just like ship it to customers too, right?
You can have like data analytics platform for it. We absolutely could, yeah. Wow, okay.
And do you guys encourage employees to do this? Because they also have their sales job, their data job. Is there a culture that encourages them to do this?
Very much so. And it's not even us having to push them. I think the types of people that Replit hires have this sort of agency in their drive.
So a lot of them... want to build things themselves. There's always that lens of how do we build something?
A, it helps us dog food, but then B, it helps us operate more efficiently as well. So our sales team, for example, has replaced a lot of their sales stack. Salesforce still is the system of record, but outside of that, it's all the tools are our own.
So yeah, it's rather common and it's amazing to see because as product builders, this is all the feedback that we get today. Yeah, I feel like this is how you become like a 10x engineer or a 10x salesperson. Like the 1x person is just thinking about doing their core job, right?
And like maybe like the 2x person is doing like building a personal tool to do their job better. But a 10x person is building a tool that the entire team or the entire company can use. Then you're basically helping everyone else, you know?
Exactly, you're creating leverage for the whole company. That is truly the 10x definition, yeah. Okay, so let me answer this.
So... What SaaS tools do you think can actually survive inside companies then? If you can just vibe code everything.
Is it like the Workdays and the Jira start too complicated to remove? Yeah. I think we were briefly touching on this earlier.
But essentially, the test, I think, is if the cost of code goes to zero, what do you pay for? And I think if you apply that test, the few things that stand that test are it's trust. It is data.
It's infrastructure. It's atoms, actual physical things. It is labor.
It is network. So I think software that offer that, where it's not just an interface over some data. Those are valuable and will continue to exist.
I think a few canonical examples here, you mentioned them. Salesforce, it is the system of record for your sales side of things. And so it should stand even though more in a headless capacity because everyone's building their tooling on top of it.
Workday stands because it is paying people as a regulated industry. You need to pay it right. They're taking on that risk.
That's valuable. You know, on the consumer side of things, a social network is rather valuable. That's not something you can buy code and do it yourself.
So a DoorDash is valuable. It is, you know, actual humans are bringing you your food. So I think that test has really stood, you know, it's helped at least me think really about, well, about what software is going to survive in this new age.
Yeah, so it's probably, it's not just the UI and the software, it's like some sort of data or like services or some other thing that's included as part of it. Exactly, yeah. And the good news is that like, because, you know, I don't want to talk too much shit, but like, I don't think anyone really enjoys using Workday, the UI, but I think all these companies are building, you know, APIs and, you know, CLIs and MCPs.
So you can just use these tools to like access all these other tools without actually having to click all the buttons. So I think that helps a little bit. It has been amazing to see that existing SaaS companies have leaned into that.
They're going headless. And then, well, now that agents can operate browsers and computers, some of that restriction goes away as well. They could reverse engineer APIs or they could take actions on your behalf as well.
Yeah, you can do that. So yeah. Okay, cool.
So the last topic I want to talk to you about is just get some hot takes from you because there's a lot of chatter right now about... open source versus closed source and app layer versus the model layer and all that kind of stuff, right? So I'll tell you, man, like for a while I thought, hey, you know, like Anthropic and OpenAI, they're building the model and the harness together.
They're going to optimize the hell out of both. And like as an app layer company, you're kind of screwed because they just want to optimize themselves, right? They want to take over the full stack.
But now there's so many awesome models out there. I feel like app layer companies are actually really in a strong position. Do you have any thoughts on this?
Because I don't think Replit actually lets you pick the model, right, that you use. Yeah, well, I love hot takes and I love this topic because I, you know, I'm right there with you. I think this is a great time to be an app layer company.
In a lot of ways, the pieces that we've always had and a lot of app layer companies have had is coming true where models are becoming commoditized. And so more and more, you know, like if you think about Replit and our sort of differentiation from these models, from these model labs, So in some ways, yes, first of all, it's the product differentiation of the last mile.
That is super important. It's, you know, cloud code giving you the code, but hosting is where, you know, and hosting in a reliable way is what matters. It's the security.
It's the distribution. It's all of that extra packaging that meets the user where they are and users paid for that. So I think while that's the core thing and that's what we're focused on a lot too, Being model agnostic is a feature itself because, you know, like we were talking about in the last two weeks with Gwen and Kimmy, the discussion has fully come to life of the latest and greatest changes all the time.
And if you're an enterprise or even a consumer, I don't know if you want to be tied to one. put all your eggs in one basket and, you know, be tied to one model app. It's the part about latest and greatest.
It's the part about cost, obviously, as well. You know, token maxing is coming to an end. And so do you really want to be beholden to one company and their pricing?
Really not. Maybe different tasks. If Kimmy can do it for one tenth, you let it do that.
And then lastly, resilience as well. a single provider can go down. We've seen that pricing changes all the time, policy changes.
It's just hard to even keep up. And I don't think enterprises or consumers appreciate that. It's almost great to have that indirection layer, that app layer that is model agnostic and that helps you orchestrate and run all your AI workflows and build yourself on top of those.
From your point of view, is it true that the fables of the world are like, do you use them more for planning? more complicated, like, I guess, re -architecture work. And then for building features or something, you just use like a open source or like, how do I think about which model to use for what?
Yeah, I think the mental model of a frontier model, like a Fable, like a Sol, like a KMEK3 to do sort of the hard part. And then an orchestrator that lets you delegate to sub -agents that are running it on cheaper models. is definitely, definitely the right approach.
And when you talk to enterprises, yeah, so they're not doing like token max leaderboards anymore, right? They actually care about cost now? That is firmly gone.
People care about, in the ROI, the R part has become very, very relevant. Got it. That's actually really hard to measure, dude.
Because like ROI depends on the function, depends on the team, you know? Because yeah, this token cost can easily go up. It's a funny meme if you've seen of lighting a cigarette with a flamethrower.
Yeah, I've seen that. All right, well, let's make some predictions. Let's close with this really hot take.
It's great if you optimize for the broader business ecosystem where American businesses get to use the latest and greatest at good prices. That means AI advances. Everyone gets to use AI.
And I think that's going to do well for us as a nation. So I'm very bullish. of the future in that sense.
All right. Well, thanks so much, man. Like, so where can people learn more about Replit and about you?
Yeah. Replit. Replit .com.
Go try it out. It is, it's one of those products. When I first tried it, it was the aha moment that took me out of my parental leave that I was on to go talk to Replit and go join him.
The aha moment is real. And, and you know, it's a great time to be an entrepreneur. or a creator, go build something.
For me, you can find me on Twitter, on LinkedIn. It's just my first name and my last name. And yeah.
Thanks to you and team for building a great product. And yeah, I'm really excited to build more stuff and turn them into real businesses. Fingers crossed.
Yeah, I hope you make some money while you're sleeping. I think you've got the audience down. Really appreciate you having me, Peter.
This was a great conversation. Thanks, Amal.
The Hook
The bait, then the rug-pull.
Before Amol Jain even says hello, the episode opens on the number that sets up everything else: a dev agency wanted $100,000 to build a certification platform a non-coder built himself in three days. What follows is Replit's head of product engineering walking through exactly how three non-technical founders turned AI-built prototypes into real six-figure businesses.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
A creator puts GPT-5.6 and Claude Fable 5 through six real build tasks -- travel sites, games, browser agents, and a nutrition tab -- to settle which one earns the daily-driver seat.
Anthropic's Thariq Shihipar walks Peter Yang through /goal, sub-agent workflows, and a planning habit built entirely around finding out what you don't know before you build.
00:28
01:09
01:34
02:01
02:55
03:27
03:55
04:17
04:41
05:11
05:29
06:05
06:32
07:03
07:19
07:45
08:47
09:15
09:42
10:09
10:36
11:03
11:30
11:57
12:51
13:18
13:39
14:12
14:39
15:07
15:34
16:01
16:28
17:22
17:49
18:16
18:43
19:37
20:04
20:31
20:58
21:26
21:44
22:26
22:47
23:14
23:41
23:55
24:38
25:02
25:17
25:56
26:23
27:04
27:45
27:59
28:36
29:06
29:33
30:21
30:43
31:21
31:39
32:15
32:42
33:10
33:47
34:02
34:22
34:45
35:50
28:29
41:48
41:17
28:52
Video of the Day21:17
Video of the Day44:39