It’s now more easier than ever to build a software business with AI. In this article, I’m going to go through my exact process & journey building a $300K+ ARR consumer app in the healthcare space. I set out to build this because:
- I was curious if the hype about vibe coding was real or fake
- I wanted to teach myself how to build an AI agent product from scratch (writing evals, picking models, building tools/MCPs, managing context, etc)
- I had a problem myself I genuinely cared about solving for a relatively untapped user segment
blip in oct when we got banned from app store for 2 weeks (oops)
I started building this product in the fall of last year and hit $300K ARR as a team of mostly myself + AI (brought on a couple more people a few months in). For context, it took me several years, millions of dollars in VC funding, and a team of very talented engineers to get to the same milestone at my previous startup.
I’m hoping that by the end of this post, you should have enough information to go launch your first app and make enough money to quit your full-time job. I learned everything I know about scaling consumer apps from other creators on X (like https://x.com/@ErnestoSOFTWARE, https://x.com/@jackfriks, https://x.com/@juliapintar ) and wanted to pay it forward to my friends who might be curious about the process. In this article, I’ll go over:
- how to pick the right idea
- validating the market first
- build & launch
- marketing the app profitably
Step 1: Picking the right idea
Now that AI has made it so easy to build an app, you should be spending as much time figuring the right things to build as you are actually building it. One way I see a lot of people fail is by picking a random idea based on their own made-up problems and then spending months obsessing over the details, and ultimately realizing that no one actually wants it.
You need to put yourself in the shoes of the average consumer. Spend time scrolling TikTok, Instagram, and Reddit in your niche. If you’re not the kind of person who consumes short-form content, you can also look at search terms on Semrush/Google Trends, join niche online communities, see YouTube popular videos, etc. Talk to literally anyone outside of tech for 5 seconds.
I picked the Health & Wellness niche because it was an area I cared about personally and had already spent a lot of time as a consumer understanding the space. I also spent time doing user interviews with other people who had the same problems as me, and I truly wanted to build something they would love.
Use tools like Astro for App Store keyword research — you want to strike a balance of there being some existing demand for your product but not too much competition (which will come in handy later for Apple Search Ads). In my case, there was only one formidable competitor at the time who was spending money on ads on every platform, but I thought their product sucked and I could build something more way more useful.
Once you’ve picked a problem space, decide the 1-2 marketable killer features that will deliver value.
Easy formula for this: pick a big successful app’s existing features and scope the marketing to a much smaller user segment (Calm for manifesting/astrology, Duolingo for learning chess, Cal AI for women etc). Don’t be afraid to copy what works, they’ve done years of iteration to perfect their strategy.
Step 2: Content comes first
Before you even start building anything, start a TikTok & Instagram account and start posting founder-POV content talking about your app. Tell your story, talk about features, practice UGC-style videos, and keep your account warm in your niche (can write more on account warming). If you’re able to consistently hit 10K+ views, you are onto something!
Build a small mockup or prototype of your killer feature and include it in your videos. Do people resonate? Are they asking for the download link? Are they joining your waitlist?
This part might take a few weeks of doomscrolling to get right. Unfortunately, you cannot delegate this work. (I can also write a longer post with video formats to try, example accounts, etc)
Step 3: Vibe code it
Before I built it, I had a huge Figma board where I screenshotted other popular apps’s onboarding, paywalls, and core feature screens. You can use tools like Mobbin and Variant for inspiration. But nothing beats actually downloading and using all of your major competitors (I probably have 100+ random apps on my phone). Also see tools like AppFigures for seeing what apps are growing fast in your niche.
pure chaos
Once you have inspiration, do a little bit of PM-ing upfront to make sure you understand your user, their main problems, and the right MVP.
Then comes the fun part: use your vibe coding platform of choice to bring it to life! Don’t worry too much about the design or polish here, your main goals should be long onboarding + hard paywall + 1-2 core features that work well. Long onboarding sounds counterintuitive, but for apps in most categories, having more screens in onboarding helps the user commit more to the app before seeing a paywall. I would also add a review popup at the end of the onboarding so you don’t get stuck with low reviews right after launch.
I built my app in Cursor + Claude Code in a few weeks (this was back when coding models weren’t as good). Nowadays, the vibe coding stage shouldn’t take longer than 1-2 weeks. If you’re spending longer than that, you’re probably getting nerdsniped.
I use RevenueCat for subscriptions and PostHog for session replays + product analytics. Both are cheap and work great.
Getting your app reviewed by the App Store can be a bit of a hassle, it took me ~1-2 weeks to get approved the first time. Use Claude/ChatGPT extensively here to go back and forth on your app description, screenshots, and screens to look for potential issues. Be patient - most get rejected on your first try.
Step 4: Marketing, monitoring, & iterate!
Once you launch, it’s time to go back to your loyal TikTok followers and give them the link to install the app. Make a Posthog dashboard to track your entire funnel: app install → go through onboarding questions → paywall conversion rates → app usage → retention. Set up Sentry to monitor for errors. Get addicted to watching session replays and fixing issues you find.
If the funnel doesn’t feel too leaky (compared to public benchmarks), it’s time to go hard on marketing!!!!
Keep posting your founder content, but now it’s time to find a few more creators and turn your videos into a content factory. Hire a few college students on SideShift and help them create NEW TikTok/Instagram accounts, warm them up to your niche, and start posting 1-2x a day using exact scripts you’ve written. A great formula here is to find videos that have already done well in your niche and find a subtle way to plug your app (screenshot overlays, feature demo, or even just add the app name to caption/subtitles). There’s great resources out there finding hooks/formats, but if you’d like a longer post on how to do UGC successfully, let me know!
Aside from UGC, the one channel I wish I launched earlier is ASA. If your app has any latent demand whatsoever, you absolutely need to be doing ASA. It’s so braindead simple to make an Ad account in ASC and start targeting keywords in your niche. Also target your biggest competitors too. You can use Claude to help you optimize the campaign once you launch, but this was such an easy way to increse our installs and I wish I had done it sooner.
Depending on your app, you can also experiment with Meta Ads, Reddit (organic + paid), SEO/GEO, Google Search ads, YouTube, FB Groups, etc etc etc. But I would argue all of this is is a distraction until you’ve actually cracked organic first. Spending too much on paid early will tank your margins and distract you from talking to your users more directly. Also, once you’ve cracked UGC, you can repurpose that content more easily later for paid.
Conclusion
The best time to start is now! It’s not too late!
If you have any other questions about how to get started, reply here and I’ll make a follow-up post :)
And if you've gotten this far, I have a big update coming soon for my next move 👀
