$2M ARR in 6 Months: Daniel Heintzman's Secrets Behind Sway AI, the App That Fixes Your Dating Profile

Sell before you build. Mass-produce slideshows. Brand every slide. Default to annual. The full teardown of how Daniel Heintzman engineered one of the fastest consumer app launches I've ever covered.

This Newsletter focuses on the overseas developers making absurd amounts of money in the AI era.

Today’s subject: Daniel Heintzman.

Daniel built an app that uses AI to analyze and optimize your dating app profile — photos, bio, everything.

Within six months of launch, he grew it to $2 million in annual recurring revenue.

Sure, it’s the kind of app that looks sellable at a glance — but $2M a year within six months is still staggering. And when I actually dug in, I found more secrets behind this growth than in almost any case I’ve covered here before.

On top of that, Daniel is now Head of Product at the mega-hit Cal AI.

Yes, that Cal AI — the app built by 17-year-old Zach Yadegari, which grew to $5M a month in under two years and it was just acquired by the giant MyFitnessPal.

Daniel actually joined Cal AI when it was doing around $1.5–2M a month, and he was a key force in growing it to $5M a month.

So today, we’re unpacking the secrets of a man who has driven outrageous revenue across multiple products.

💘 An AI App That Boosts Your Match Rate

Let’s jump straight into the app that hit $2M ARR in six months.

It’s called Sway AI.

Upload screenshots of your profile on Tinder, Hinge, Bumble, or any dating app, and the AI critiques your photos and bio — then tells you exactly what to change to get more matches.

According to Daniel, the idea was sparked by two apps that were growing explosively at the time: RizzGPT (which generates conversation lines) and Umax (which scores your face).

Both of those apps were built by Blake Anderson, who, starting right after ChatGPT launched, took consumer AI apps viral one after another and at one point was earning around $600K a month.

Daniel was watching Blake’s run from the sidelines.

Around that time, clones of Blake’s apps started flooding the market. Apps I’ve covered here like FaceKit and Glam Up are exactly that — copies of Umax’s concept.

But Daniel took a different angle. Instead of copying either app, he thought: why not combine them?

That’s how Sway AI was born — an AI that optimizes your dating profile photos and bio in one place.

📲 Sway AI Is Obsessed With Onboarding

Let’s dig a little deeper into the app itself.

Daniel says the thing he obsessed over most while building it was onboarding.

For him, onboarding is the sales funnel itself. If you want a profitable app, onboarding is everything.

He breaks effective onboarding into these steps:

Make users aware of why they need the app
Make them imagine the future the app unlocks
Personalize the experience through questions
Build a thoroughly smooth, low-friction path

Let’s look at how Sway AI actually implements this.

First, a screen that makes it clear why you need the app and what it can do:

Then it shows you the “ideal outcome.” What happens if you use this app — and what happens if you don’t:

Then it personalizes the experience through questions:

And finally, right before the paywall, it hits you with the “ideal outcome” one more time:

Here’s the point Daniel stresses hardest: before asking any questions, clearly explain what the app is and what users are paying for.

At first, they skipped this explanation and jumped straight into questions. The result? A flood of support emails: “So what is this app, exactly?” “What am I even paying for?”

When you ask questions without context, users think: “Why am I entering all this information?”

So: present the problem first, show how the app solves it, hand users the context — then ask questions. Just fixing this order transformed the entire experience.

One more tip on question design. Questions matter for personalization, but the key is minimizing friction. Never make the user think.

Every choice is multiple-choice — just tap.

The moment you add a free-text box that says “describe your struggles,” users abandon in droves.

The goal of onboarding is to never stop the user — to carry their momentum smoothly all the way to the paywall.

I’ve felt this firsthand. I once introduced a screen in my own app that made users hesitate a little. I thought it added value — and conversion dropped dramatically.

Longer onboarding tends to convert better, but longer isn’t automatically better. The slightest friction along the way and users bail.

You have to achieve a frictionless, seamless flow — relentlessly. Walk through your own onboarding again and again, and have friends try it in front of you.

I look at app onboarding flows every day, and I can’t overstate how much they matter. When you see a great screen in another app, steal it. But the optimal onboarding differs by app — so test properly, and roll back what doesn’t work.

If you want to study onboarding screens, check out Onbo Hub, the tool I’ve been steadily building. Sway AI’s full onboarding flow is on there too.

So now we have a decent picture of what Sway AI is. But how did they actually sell it?

Again — $2M ARR in six months is a phenomenal number.

Here’s the fascinating part: they didn’t start by building the app.

They got users to pay money before the app existed. I’ve covered demand-validation-before-building approaches before, but this case is special — and surprisingly replicable. Let’s get into exactly how they did it.

🍳 Sell Before You Build: The Ultimate Marketing Move

So the concept was set. What did Daniel do next?

As I said — he didn’t build the app.

Instead, he and his friends launched an Instagram account.

The content? Posts like “how I cook on Hinge” — pickup lines and profile banter, presented in a funny slideshow format:

https://www.tiktok.com/@swayai.app/photo/7363685476093447455

That post pulled over 800,000 views.

The app didn’t exist yet, so there was no app promotion anywhere. They were simply testing whether they could go viral on social — posting content in every format they could think of.

Up to this point, it’s a fairly common story. Here’s where they got really smart.

They went after the real question: will users actually pay money for this? — before the app was finished.

Specifically: a few weeks into posting, they told their followers:

“DM me and I’ll review your profile. Send a screenshot.”

And when the screenshots arrived, the reply was:

“That’ll be $15.”

And people paid. With the app not existing by even one millimeter, people were paying money for profile feedback.

Note what they didn’t do: no dedicated promo video, no hard sell. Just a CTA slipped naturally into the captions of their regular posts. The vibe of the account stayed intact.

I think this is a beautiful move.

Most people think “I need to build an MVP first,” spend ages building, launch — and then despair: “wait, nobody’s paying…”

Daniel’s team did the reverse. Before investing the effort to build an app, they took real money, manually. If people pay $15 here, that’s the strongest possible proof the problem is worth paying for.

And if nobody pays? You just saved yourself from building the app at all. No wasted months.

I’ve covered plenty of cases of validating demand before building a product. But actually charging money at that stage is rare. And without charging money, you never truly validate demand.

I’ve also covered the approach of asking people to pay upfront — “I’m thinking of building this, want to pre-pay?” But that pattern is brutal. You have to go out and sell, aggressively.

Daniel’s pattern, by contrast, flows beautifully. Instead of showing mockups of a future app, they delivered the solution itself from day one — like a dating profile consulting service.

And once it worked, all they had to do was replace the human with an app.

Genuinely elegant.

This approach can work for almost any app theme. The core value of most apps can be delivered as a human-powered service first. Before you write code, try selling the service — and see if anyone actually pays.

🎞️ Low-Cost Slideshows That Rack Up Million-View Hits on Repeat

Having validated demand with a $15 manual service, they launched the app — and kept testing short-form content formats.

They discovered that slideshow-style content was by far the most cost-effective way to go viral.

Slideshows get surprisingly high view counts, and they’re easy to mass-produce.

Daniel’s team studied the slideshow formats that were already blowing up — and copied them relentlessly.

They paid special attention to men’s self-improvement accounts, taking the copy from the most viral slides and adapting it for their own app:

They didn’t write anything from scratch. They took the exact wording of proven viral content and remade it for Sway AI.

Eventually, they landed on their winning format: slides with nothing but a text hook on Sway AI’s brand color:

Absurdly simple, right?

And that’s the point — this content is trivially easy to produce. Drop text onto a fixed brand-color template. Done.

They managed the slide copy in a spreadsheet and auto-populated the templates. Each slide took seconds to generate.

The result: a full month of content, batch-produced in one or two hours.

Once they saw it working, they flooded the zone with this format and stacked up viral hits:

“Grinding out a viral video every single day” is the biggest bottleneck in this kind of marketing. Daniel turned it into an assembly line.

I use short-form video for my own app growth, and the lesson holds: pick themes with effectively infinite content ideas that require no creative agonizing.

Look for the intersection of “themes that go viral easily” × “themes that never run out of material and cost nothing to produce.”

But low production cost isn’t even the best part of this format.

Did you catch it? The other killer detail: every single slide carries the “Sway AI” name and pitch.

The standard slideshow playbook is “app promo on the last slide only.” But then, even when a post goes viral, most viewers never see the app — they swipe away before the end.

So Daniel subtly branded every slide, imprinting the app on everyone — including the people who bail halfway through.

But — and this matters — they didn’t lead with heavy branding.

Right after creating the account, they deliberately posted polished, aesthetic carousels with zero branding. A brand-new account that’s aggressively branded from day one smells like spam, gets suppressed by the algorithm, and struggles to grow. So for the first few weeks they focused entirely on “warming up” the account — earning trust and views with harmless, high-quality content.

Only after several posts went viral and the account was warm did they roll out the orange brand color and CTAs in earnest.

With that, the machine was ready: massive impressions and conversions, at near-zero cost.

By the way, one post in this format, published on Christmas, cleared 4 million views:

https://www.tiktok.com/@mendatingadvice/photo/7452385767290178822

Christmas and dating-app marketing are a perfect match. Single people sitting at home scrolling, wishing they had someone — of course this content goes viral then. It’s almost inevitable.

Riding organic content alone, Sway AI reached $30–40K a month.

…and then hit the organic ceiling.

When you’re posting everything yourselves, content volume has a hard limit. So how did they climb from there to $2M a year?

📈 Smashing Through the $30K Ceiling With Influencers

From $30–40K a month on organic, they moved into influencers and paid ads.

But their approach was a little unusual.

The obvious play would be boosting their own viral organic videos as ads. They didn’t do that.

Instead, they partnered with a single influencer: Defund Simp, a creator in the dating-and-men’s-advice niche.

They had him produce dedicated ad creative, and ran that as paid ads.

But here Daniel hit a structural problem: influencer marketing is fundamentally hard for niche dating apps.

Why?

Because big dating-niche creators almost all sell their own coaching or courses. Their business is “I’ll give you feedback on your profile and photos.” And Sway AI automates exactly that with AI.

Promoting Sway AI means cannibalizing their own product. So the bigger the creator, the less likely they are to work with you.

Now compare that with Cal AI, where Daniel is now Head of Product.

Cal AI is a calorie-tracking app. Fitness influencers sell diet programs and training courses — “automatic calorie counting” doesn’t compete with their business. That’s why influencer recruitment is so much easier there.

The lesson from these two apps:

Pick a niche where your app can coexist with influencers’ existing businesses.

Despite the influencer struggle, they made the ads work and kept scaling.

💳 Defaulting the Paywall to Annual to Push Up LTV

Revenue isn’t only about marketing. The paywall itself has machinery inside.

Sway AI defaults to the annual plan.

Many of today’s top-grossing apps show a weekly plan next to an annual plan, make the annual look dramatically cheaper, and steer users toward it.

Why annual?

Annual plans push LTV higher — and the cash lands upfront, which you can immediately reinvest in growth.

They tested weekly plans too, but growth was dramatically slower. And from studying other apps, Daniel reached the same conclusion: annual-default apps consistently show higher LTV.

💰 The Add-On Purchases Generating 11% of Revenue

Sway AI’s monetization doesn’t end with subscriptions. Arguably, this is where it gets interesting.

There are two layers of additional purchases stacked on top.

Layer one: extra profile-scan credits.

After launch, they noticed users running the scan flow far more times than expected — so they productized it as an add-on purchase. This alone now accounts for about 11% of revenue.

Layer two: photo packs.

Upload your face, and the AI generates flattering portraits of you. Pricing: $14.99 for one pack, $49.99 for ten.

Here’s the key property: photo packs have no spending ceiling. So “whale” users emerge naturally — people who just keep buying, without limit.

According to Daniel, this is exactly the structure that makes Tinder, Hinge, and Bumble so profitable: the top 1% of users generate the majority of revenue.

“People will pay endlessly to look better.”

AI happens to be exceptionally good at this — and it delivers an instant “wow.” Pointing unlimited purchases at this universal human desire is Sway AI’s revenue engine.

Short-form marketing and ads bring the users in; the monetization design raises revenue per user. That’s how Sway AI reached $2 million a year.

🧭 Daniel’s Framework for Finding the Next Hit App

Drawing on his Cal AI experience, Daniel has articulated what makes a good app idea — and it’s remarkably practical.

His three conditions:

Simple enough to understand instantly in a marketing video
Novel
In an already-large market

Take Cal AI: “snap a photo, get the calories” is understood the moment you see it in a video. Nobody was doing it at scale at the time. And calorie tracking was a massive market with a giant incumbent — MyFitnessPal.

All three conditions, perfectly aligned.

Ideally you want all three — but beware the classic trap: chasing novelty (condition 2) so hard that you drift out of the large market (condition 3).

The two can look like a trade-off. Daniel’s advice: start from a large market where demand is already proven.

The right answer is to take a proven, large market and add exactly one novel angle. Cal AI added “measure by photo” to the tired old market of calorie counting. That’s the template.

One more idea: plug an app into content that’s already going viral.

Find a trend that’s already blowing up on social, and slot an app into it. Even if you can’t validate your idea against existing apps, the fact that the content is viral is itself a demand signal.

📝 How to Use This Case — and Wrap-Up

So that’s Daniel Heintzman and Sway AI.

This one taught me a lot.

First, the onboarding insight — show the app’s value before asking questions — hit home personally.

I had actually designed my own onboarding as “questions first, value after,” and conversion never impressed. I’d even started thinking the value-demonstration step was a waste. Turns out I may have had it exactly backwards. That reversed flow is absolutely worth testing.

Second, their launch approach is brilliant:

Don’t build the app first. Verify that people will actually hand over money.

This applies to so many apps. The core function of most app ideas can usually be delivered as a human-powered service without any software. Before you build, try offering the service — and see if real money shows up.

And once again: slideshow-style short-form content is strong. I watched this case prove that a sharp enough hook can reliably manufacture virality. If you’re struggling with short-form, give slideshows a real shot. I’m recommitting to them myself.

So — keep experimenting, keep building.

Let’s push forward together. In this AI era, one person can do anything. What a ridiculously fun time to be building.

Let’s get to work!

That’s it for this deep dive into Daniel Heintzman.

Thanks for reading to the end! If anything caught your attention or you have questions, just reply to this email.

And if you mention my X account with your thoughts, it’ll make my day — I always respond!

See you next time!