Daniel Kwon - He made $2.2M from building simple apps. Here's how (no-code)

Found a video already going viral for other AI writing tools (a kid getting caught in class, ~10K likes, poor quality, no on-screen text). Swapped its sound for audio from a different clip of a teacher yelling 'No AI on this test' so it felt like a real student getting caught, added Snap-style captions ('How the f is this kid acing every exam?'), then zoomed in to show he was using the AI website: a straight hook then demo. Posted by the guest as a faceless solo creator on the brand's TikTok; it was then reposted on X by random accounts. He calls this 'Frankensteining': find videos that already went viral on your subject and stitch them together. Aim for one extreme: either obviously a cringe homemade ad or something that feels completely real; lower quality leans into realism.

Content leaned into the frustration the guest felt as a student: too much busywork that doesn't lead to a job. Example: 'ChatGPT just broke university' with a voiceover saying half the assignments students get are just busywork. Pick the side with the larger or louder emotion in an outraged group and act as one of that crowd.

The guest, a former influencer, made the videos as the solo, faceless, founder-led UGC creator for the brand account. Earlier, 1-2 weeks of videos for the predecessor product (Pandora) got ~100K views but few conversions; the pivot to the bypass tool is what converted.

Tested paid ads; no details on platform or spend.

The team was building Pandora (ChatGPT for Google Docs). When GPTZero launched, they decided the same day to build a bypass tool; a co-founder prototyped it in Figma that night.

Newer videos were made by UGC creators who filmed natively in the Snapchat app with Snap-style captions. Lower quality is fine: it leans into the 'realistic' pillar, since viewers recognize the frame rate and pixelation of Snap stories. The guest notes Snap captions were in style for a while but less so now.

The website used several schools' logos, and the founders listed Conch in their LinkedIn bios instead of building in stealth. Local news covered high school students using Conch to bypass AI detection.

Talked to three or four potential buyers; interest came from the margins and numbers. Some were competitors who pretended to want to acquire them and just took information.

Script: 'We made way too much sour candy, so we're putting it on sale. Not the typical junk you see filled with sugar and chemicals, but real fruit covered in real sour. Just look at the ingredients list. Some of you guys are snagging these 18-pack variety boxes for less than $20, so this definitely isn't going to last much longer.' Shot scrappy, like it was made in the founder's kitchen. Every UGC creator brought on was given it to rinse and repeat, mixed with other narratives such as Keith Lee reviewing the product.

One of the accelerator's content pillars for its brands: focus on clearance of products. Putting a flash sale on an item boosts both the product and the brand into For You pages; the content then has to convert that click-through.

The 'small business factor' was a major pillar across the accelerator's brands: even though FinalBoss Sour is VC-backed and makes millions a month, content feels shot in the founder's kitchen or bedroom, building a cult around one person. The guest says the scrappy style that worked for Conch worked even better for TikTok Shop brands; the best affiliate videos read like Amazon reviews.

Product hero images give an instant sense of value (how much is in each box). The brand gamifies the product with limited drops and a character associated with the brand, similar to a streetwear brand.

Anime fans were already posting 'We need the system in real life', so the demand existed. Posted carousels about fans being excited that the system could exist in real life, timed with the Solo Leveling season 2 finale, and milked that one format; then partnered with influencers.

Partnered with influencers and fandom meme pages, paying roughly $1-2 CPM (deals varied). Revenue ran about $6-7 per 1,000 views.

Fans already wanted the app and a few existed, but none were optimized: no long onboarding flow, no social features, not built to convert, mostly monetized with ads. The guest built a converting version with a long onboarding flow and calls ads 'the worst way to monetize your app'. His framework: find or build a 'cult' riding a momentum wave; building the cult is better for LTV.

'Frankensteined' videos on the app's own page: podcast clips of Christian influencer Bryce Crawford plus hooks farmed by searching the phrase 'Bible reading', each followed by a standard app demo. Some also ran on creator accounts, but the most viral were on a new Instagram account (guest thinks Instagram is easier to grow from scratch). Organic was later paused because the team didn't want such a meaningful app to depend on hacky UGC.

Borrowed Duolingo-style micro-animations and art style while Duolingo was being criticized for its AI-first push, framing the app as 'God-first lamb over the AI-first owl'. The launch push built a Discord community.

Briefly added an AI chat feature; members in the Discord pushed back because the launch content was anti-AI. Removed it and turned it into a feature that generates a custom devotional, with no mention of AI in customer-facing copy.

About a month before recording the co-founder halted all creator ops to focus on product: tightening social loops so users invite prayer and accountability buddies for daily Bible reading, and spreading through schools and churches like NGL and Gas did.

Ideated with AI and a few calls, built in two weeks alongside a two-week growth sprint; nothing was optimized and the app broke a lot.

To find content-market fit: post about your product yourself on TikTok and Instagram and gather data on what content works, rather than hunting for one format. Search phrases loosely related to the problem or community and stitch those videos. Work with influencers and convert them into full-time UGC creators who run an account posting only for your brand. Don't touch paid until you've cracked at least five winning formats.