Retora presents a novel intersection between the $150 billion mobile ad market and the $100 billion AI training sector. The company proposes an 'Offerwall meets Data Training' model, where mobile gamers perform RLHF tasks—such as identifying AI-generated images—in exchange for in-game currency. This approach seeks to solve the labor shortage in high-quality AI training data while providing developers with revenue that exceeds traditional ad payouts. With a team led by a Forbes 30 Under 30 founder and a Zynga veteran, Retora leverages its existing 2 million organic downloads from previous titl…
Key takeaways
- Retora identifies a critical bottleneck: AI companies need human feedback, but demand has outpaced the available workforce (Slide 2).
- The business model functions as a three-sided marketplace where AI companies pay for data, developers get a higher cut than traditional ads, and Retora takes a commission (Slide 6).
- Market tailwinds are strong, with AI projected to reach $1.8 trillion by 2030 and mobile ads hitting $621 billion by 2029 (Slide 7).
- The company has historical traction, citing over 2 million organic downloads for their 2015 release 'Merchant' (Slide 9).
- Retora differentiates from competitors like Scale and Sama by using a crowdsourced gaming audience rather than underpaid direct labor (Slide 11).
- The team includes David Bettner, founder of Newtoy Inc and a former Studio Director at Zynga (Slide 12).
- The $2M Seed Round is specifically earmarked for SDK development, server infrastructure, a dashboard, and partner acquisition (Slide 13).
- The deck uses a specific 'Merchant Guilds' sequel launch in Q3 2023 as the primary test pilot for the delivery mechanism (Slide 10).
Retora Pitch Deck Teardown
Slide 1: Title and Value Proposition
The deck opens with a clear, high-contrast title slide. The company name, Retora Games , is paired with a succinct sub-headline: "Providing ML training data through mobile games." This immediately positions the company at the intersection of gaming and artificial intelligence, avoiding the common pitfall of being too vague about the industry sector.
Slide 2: The Problem
Slide 2 identifies a supply-chain bottleneck in the AI industry. It states that "AI companies need human feedback to improve their products... but the demand for the data has grown beyond the available workforce." This sets the stage for a scalability solution, implying that current manual labeling methods are insufficient for the current pace of AI development.
Slide 3: The Solution
The solution is presented as a hybrid model: "Offerwall Meets Data Training." This slide is minimalist, serving as a transition to explain how the mechanics of mobile game monetization (offerwalls) can be repurposed for a more lucrative end-market (ML training).
Slides 4-6: The Business Model and Mechanics
These three slides walk through the value chain. Slide 4 explains that Retora provides AI/ML companies access to a large user base via mobile games. Slide 5 details the user incentive: "We incentivise these users to provide feedback by offering in-game rewards." This is a critical distinction from traditional labeling; the 'pay' is digital currency, which has a lower marginal cost for the developer but high perceived value for the player. Slide 6 illustrates the financial flow, noting that AI companies provide a "higher payout than a traditional ad" and that Retora takes a cut of the transaction between the AI company and the game developer.
Slide 7: Market Opportunity
Retora highlights two massive, growing markets. Artificial Intelligence is cited as a "$100 billion Current Market" with an estimated growth to "1.8 trillion by 2030" at a CAGR of 32.9% . The Mobile Ads market is listed at "$150 billion" currently, growing to "621 billion by 2029" at a CAGR of 23.2% . By positioning themselves in the middle, Retora suggests they can capture value from both sectors.
Slides 8-10: Traction and GTM Strategy
Slide 8 mentions a 2022 test called "TBDNE" (This Bird Does Not Exist) which demonstrated that audiences enjoy the labeling process. Slide 9 provides historical credibility, noting that in 2015 the team released "Merchant" and achieved "over 2 million organic downloads." Slide 10 connects these dots to the future, announcing "Merchant Guilds (Q3 2023)" as the initial test pilot where they have full control over the delivery of the RLHF tasks. The slide includes a mockup showing a player choosing between four AI-generated wizard character designs to receive 5 gems.
Slide 11: Competitive Landscape
The deck categorizes competitors into two groups. Offerwall polling services (Tapjoy, Pollfish) are dismissed for having "poor reputation among users" and no AI solutions. AI RLHF companies (Scale, Sama, Surge) are described as having "underpaid workers and no crowdsource solutions." Retora’s implied advantage is a happier, more scalable workforce (gamers) and a more specialized technical offering than generic pollsters.
Slide 12: The Team
The team slide features Tyler Coleman , the founder and a Forbes 30 Under 30 honoree, and David Bettner , an early investor and advisor who was a Studio Director at Zynga and founder of Newtoy Inc (the creators of Words With Friends). The slide also lists four staff members—Denis Krasakov, Viacheslav Bogomazov, Garret Wied, and Clay Burton—represented by pixel-art avatars consistent with their game 'Merchant.'
Slide 13: The Ask
Retora is seeking a "$2M Seed Round." The funds are allocated to four specific pillars: SDK, Server, Partner Acquisition, and Dashboard. This indicates a plan to move from a single-game pilot to a platform model where other developers can use Retora’s infrastructure to monetize their own games via ML tasks.
Slide 14: Contact
The deck concludes with a simple contact slide featuring Tyler Coleman's email address.
What Works in the Retora Deck
Clear Market Convergence: The deck does an excellent job of explaining why two disparate industries (Mobile Gaming and AI Training) belong together. The logic that a gamer’s time is cheaper to 'buy' with virtual currency than a professional labeler’s time is with cash is a compelling economic argument.
Proven Developer Pedigree: Citing 2 million organic downloads for a previous title ('Merchant') is significant. It proves the team isn't just theorizing about games; they have built and scaled a successful product in a crowded market. The inclusion of David Bettner adds substantial industry weight, given his history with Zynga and Newtoy.
Specific Use Case: Slide 10 provides a tangible example of what the product actually looks like. Showing a player choosing a character design for gems makes the abstract concept of 'RLHF' immediately understandable to a non-technical investor.
What is Missing from the Retora Deck
Unit Economics: While the deck claims that AI companies provide a "higher payout than a traditional ad," it lacks specific figures. Investors would want to see the delta between a standard rewarded video ad eCPM and the projected eCPM of an RLHF task. Without these numbers, the 'higher payout' remains a hypothesis.
Data Quality Assurance: A major concern in crowdsourced labeling is 'garbage in, garbage out.' The deck does not explain how Retora ensures that a gamer clicking buttons for gems is providing accurate data. There is no mention of consensus algorithms, trap questions, or verification layers to prevent low-quality inputs.
Regulatory and Ethical Considerations: As AI training data comes under increased scrutiny regarding copyright and labor practices, a slide addressing how Retora handles data privacy or the ethics of gamified labor would have been timely.
What a Founder Should Copy
The 'Offerwall' Analogy: If you are building a complex B2B product that relies on a B2C audience, find a familiar mechanic to compare it to. By calling it an 'Offerwall,' Retora uses a term every mobile investor understands to explain a new AI concept.
Visual Consistency: The deck uses the same color palette and pixel-art style as the company's games. This creates a cohesive brand identity and reinforces the idea that the founders are, first and foremost, game developers who understand their medium.
Phased GTM: The deck outlines a clear path from 'Test' (TBDNE) to 'Pilot' (Merchant Guilds) to 'Platform' (SDK development). This phased approach reduces the perceived risk by showing that the company is validating its tech on its own properties before asking others to adopt it.
Frequently asked questions
- What is the core problem Retora is solving?
- Retora addresses the scarcity and quality of human feedback required for Reinforcement Learning from Human Feedback (RLHF). As AI models grow, the demand for human-labeled data has exceeded the capacity of traditional workforces. Retora solves this by tapping into the massive, underutilized time of mobile gamers, turning leisure activities into productive data labeling sessions.
- How does Retora's revenue model differ from traditional mobile advertising?
- Traditional mobile advertising relies on impressions or clicks, often yielding low eCPMs for developers. Retora claims that AI companies are willing to pay significantly more for high-quality training data than advertisers pay for a standard ad. Retora facilitates this transaction, providing developers with a higher payout while taking a cut of the transaction fee.
- What evidence of product-market fit does the deck provide?
- The deck cites two main points of traction: a 2022 test project called 'TBDNE' which proved audience engagement with the labeling process, and the 2015 release of 'Merchant,' which garnered 2 million organic downloads. These figures suggest the team can build games people want to play and that those players are willing to engage with the data-labeling mechanics.
- Who are the primary competitors and what is Retora's edge?
- Competitors fall into two camps: offerwall services like Tapjoy and Pollfish, and AI labeling firms like Scale, Sama, and Surge. Retora argues that offerwalls have poor user reputations and lack AI solutions, while labeling firms rely on underpaid workers and lack the scale of a crowdsourced gaming audience.
- What are the specific technical goals for the Seed Round?
- The $2M Seed Round is focused on infrastructure. Specifically, the funds are allocated to developing a software development kit (SDK) that other developers can integrate, building out server capacity to handle data processing, creating a dashboard for AI clients to manage tasks, and funding partner acquisition to grow the network.
