How Dillion bootstrapped an AI photo app to $300k MRR in 45 days

Executive overview

After 18 months of failed products and almost no money left, Dillion and his co-founder had one last shot. They built Halo AI — an AI photo editor — and used mass UGC to drive 1.2 billion views in 120 days.

The model is simple: revenue = views × conversion rate. Maximise both variables and revenue follows. A single viral TikTok format, 85 creators posting 300 videos a day across four platforms, and a milestone-based pay structure did the rest.

One repeatable format plus a scalable creator system beats complex multi-angle UGC campaigns.

The core equation

• Total views = creators × videos per creator × posts per video × views per post
• Conversion rate for a given format stays constant — double views, double revenue
• Platforms don't penalise new accounts; any account can go viral on the first post
• Posting the same video across TikTok, Instagram, YouTube Shorts, and Facebook multiplies views with no extra creative work

Finding and vetting creators

• Post job listings in Reddit UGC communities and Facebook groups (hundreds of groups, thousands of members each)
• Screen for: past UGC or app/tech experience, US-based or US VPN
• Run a paid 5-day trial: $20 per video, 5 videos = $100 trial cost
• Pass criteria: consistency (posts daily), communication (proactive in Discord), understanding of viral structure (hook → build-up → payoff)
• Don't set hard view thresholds for trial — algorithm is a power law; assess editing instinct instead
• Filter for native English speakers and creators who match the target demographic

The one format

• Hook: absurd AI-generated image paired with an attention-grabbing caption (e.g. "pranking my dad I cooked his koi fish")
• Demo: screen recording of the app generating the image, with the Halo AI logo prominent on the loading screen
• Story: a text message exchange that builds tension and ends with a twist or payoff
• No deviation from this format — every creator posts the same structure
• ~5% of videos exceed 100k views; ~1% exceed 1 million views

Scaling the creator system

• 85 creators, uncapped posting (some post 20+ videos per day across all platforms)
• When one video goes viral, an automation sends it to every creator in Discord — all 85 recreate it immediately, compounding the spike
• An OpenAI custom agent trained on all million-plus-view videos generates new script ideas and hooks on demand
• The agent connects to the TikTok API via Scrape Creators to pull trending keywords and hashtags, then remixes them into new video concepts
• Community matters: creators share wins, give feedback, and coach each other in Discord voice chats — this drives consistency beyond pay alone

Creator pay structure

• Four options tested: fixed retainer, CPM, mixed (base + CPM), milestone (base + bonuses)
• Fixed eliminated — no incentive to maximise views
• Gave 30 creators a choice between CPM and milestone; 29 chose milestone even after being told CPM paid slightly more
• Milestone structure: $20 base per video + bonuses at 20k, 100k, 500k, and 1M views (~$60, $200, $500, $800)
• Effective CPM is below $1 because views in between milestones are unpaid
• Only one platform post needs to hit the milestone per video — non-stacking across platforms
• Operationally simpler than CPM: no tracking partial view counts or retroactive true-ups

Paywall and monetisation

• Default paywall: weekly subscription at $8.99, trial to convert
• Two-paywall onboarding outperforms a single hard paywall: soft paywall first (skippable) → user browses the dashboard → hard paywall triggers when they attempt to submit a prompt
• Skippable first paywall lets uncertain users see the app's value before being asked to pay
• The dashboard replicates the viral TikTok experience — example outputs across hairstyles, decor, tattoos — keeping intent high before the hard gate
• Sign-in screen placed before the paywall to link users to the database

Unit economics and growth

• Took a $100k personal loan to pay early creators; scaled at break-even for the first ~4–5 weeks
• 6:1 ROAS on early UGC spend was the conviction trigger for the loan
• Plateau at $300k MRR was forecasted using the views × conversion equation
• Current margin: ~50% ($150k/month profit)
• Next target: $1M MRR by June, using paid ads with existing viral formats as creative
• Reinvesting all profit into growth

Lessons

• Make hard decisions (e.g. dropping underperforming creators) as soon as the gut signals a problem — delay compounds
• Consumer apps offer faster feedback loops and lower capital requirements than B2B
• Focus future products on health, wealth, or relationships — higher retention, LTV, and revenue than pure entertainment