Halo AI hit $300K a month in 45 days with one video format

Scaling to $300k MRR via UGC repetition

Dillion Verma, the founder of Halo AI, achieved a rare feat in the consumer app space: bootstrapping an app to $300,000 Monthly Recurring Revenue (MRR) in just 45 days. This growth was the product of a highly disciplined, repeatable marketing engine built on User Generated Content (UGC).

Verma's success came from a highly disciplined approach to content rather than a wide variety of creative angles. While many brands attempt to manage dozens of different content "vibes" to see what sticks, Halo AI focused on one specific framework. This single format generated 1.2 billion views over 120 days.

This strategy turned social media into a predictable top-of-funnel engine. By identifying a format that maintained a constant conversion rate, the team moved away from "guessing" what might go viral and toward a mathematical approach to growth.

Context: This analysis is based on an interview with Dillion Verma, founder of Halo AI. The growth described follows 18 months of failed product attempts and was kickstarted by a $100,000 personal loan to fund the initial creator scale.

The winning creative formula: Hook, Demo, and Payoff

The reason Halo AI was able to scale so effectively is that they moved away from creative experimentation and toward a rigid, three-part framework. Verma notes that once they identified a format where the conversion rate remained constant, they refused to deviate from it.

The "Golden Goose" format consists of three distinct movements:

1. The Absurd Hook: Every video begins with an absurd image or a high-stakes premise designed to stop the scroll. For Halo AI, this often involves an AI-generated image that is intentionally provocative or strange (e.g., "Pranking my dad I cooked his koi fish").
2. The Tool Demo: Immediately following the hook, the video demonstrates the app in action. This includes a screen recording of the AI generation process and a prominent loading screen featuring the Halo AI logo. This serves as a subtle but repetitive call to action.
3. The Storytelling Payoff: The final segment uses a text-message conversation format to build tension. By showing a "story" (like a prank playing out via text), the video makes the viewer wait for the resolution, increasing engagement time and the likelihood of virality.

By enforcing these constraints, the brand ensured that even with 85 different creators, the core message and "vibe" remained identical. This lack of deviation allowed the team to treat content production as a manufacturing process rather than an artistic one.

Creator incentives: Why milestones beat CPM

Managing a fleet of 85 creators requires a pay structure that aligns their interests with the app's growth. Verma experimented with four different models: fixed monthly rates, Cost Per Mille (CPM/pay per thousand views), mixed models, and milestone-based payments.

The result of his experiment was a significant surprise. Out of 30 creators, 29 chose the milestone-based structure, even when explicitly told it might result in lower total pay than the CPM model.

Verma found that creators preferred the "base plus milestone" model because it provided the psychological security of a guaranteed floor ($20 per video) while offering the excitement of "gaming" the system to hit large payouts.

For the business, this model is highly efficient. It is operationally easier to track whether a video hit 1 million views than to calculate exact, fluctuating CPMs for every single post. Furthermore, it keeps the effective CPM lower for the brand because the bonuses are only paid when significant value (views) is actually delivered.

The AI-driven script factory: Using OpenClaw

To move from manual management to a "content machine," the team developed an AI agent called OpenClaw. This tool solved the primary problem of generative AI: knowing what to generate that will actually work.

The workflow for the OpenClaw "script factory" works as follows:

1. Data Collection: The team identifies all videos that achieved over 1 million views.
2. Transcription: These videos are fed through Gemini to extract full transcripts and a structural understanding of the content.
3. Local Storage: The data is stored in local text files that act as a "knowledge base" for the agent.
4. Inference & Generation: When a creator needs a new idea, they query OpenClaw. The agent pattern-matches the successful historical data to generate new scripts that follow the proven "Hook, Demo, Payoff" structure.

Want the technical picture?

OpenClaw is essentially a RAG (Retrieval-Augmented Generation) system tailored for social media. By connecting the agent to the TikTok API, it can also ingest trending keywords and hashtags. This allows the agent to suggest how to "remix" a proven script structure with a current trending topic (e.g., applying the "Koi Fish" prank format to an Olympic-themed trend), ensuring the content remains fresh without losing its conversion power.

This automation turns the role of the founder from a "creative director" into a "systems architect." Instead of coaching 85 people individually, the founder maintains the agent that coaches them.

The math of growth: Revenue as a predictable equation

One of the most important shifts in Verma’s approach was treating marketing as a math problem rather than a creative one. He simplified the entire user acquisition funnel into a single, predictable equation:

Total Revenue = (Total Views) × (Conversion Rate)

If a million-view video results in 500 downloads, and those users pay $50 a month, the revenue becomes a predictable outcome of view counts. This clarity allowed him to calculate a Return on Ad Spend (ROAS) of 5:1 or 6:1 in the early days. This high ROAS was the primary driver for his decision to take out a $100,000 loan; he knew that for every dollar spent on creators, he was getting multiple dollars back.

The business maintains approximately 50% margins by avoiding the high costs associated with traditional influencer marketing. In the influencer model, brands often pay upfront fees to celebrities. In the Halo AI UGC model, the brand only pays the base rate and performance bonuses, which keeps the cost per acquisition (CPA) lower.

The team currently follows a reinvestment strategy, funneling all profits back into growth to reach a goal of $1 million MRR. Additionally, there is uncertainty regarding whether these 50% margins will hold when transitioning from UGC to more expensive, traditional paid Meta ads.

Lessons for consumer AI builders

For developers and founders building consumer AI applications, the Halo AI case study offers several tactical shifts in strategy:

Prioritize "Winning Formats" over "Creative Variety" Most founders believe they need to test dozens of different content angles to find success. Verma’s experience suggests the opposite: find one format that converts, and then refuse to deviate. Once you find a high-converting "Hook-Demo-Payoff" loop, your job is no longer to be creative, but to scale the repetition of that loop.

Automate Content Intelligence with AI Agents Don't just use AI to write generic text; use it to analyze your own success. Build a system that ingests your most successful historical data (transcripts, timestamps, engagement metrics) and uses an LLM to generate new iterations. This turns "content creation" into a data-driven manufacturing process.

Optimize the "Soft Gated" Paywall Verma’s testing revealed a powerful conversion tactic: the combination of a non-gated onboarding and a gated action.

• Step 1: Allow users to enter the app and "poke around" (non-gated). This builds trust and shows value.
• Step 2: Trigger the hard paywall only when the user attempts a core action (e.g., clicking "Submit" on a prompt). This "soft" entry reduces friction and catches users at the moment of highest intent.

Watch the Retention Trap There is a fundamental tension between "viral" content and "retained" users. Halo AI is currently an entertainment-focused app, which is excellent for viral growth but can lead to lower long-term retention. To build a sustainable business, builders should aim to eventually transition from "entertainment" hooks to "core problem" solutions in high-value niches like health, wealth, or relationships.