Training Social

VISION

Become the world’s largest, most diverse, and ethically verified source of labelled video training data built through play. Turn everyday human curiosity into a worldwide-scale data engine that teaches machines to see the world as people do.

MISSION

Build a viral, challenge-driven video game that produces millions of hours of verified, high-quality video footage annually collected by game players completing real-world capture quests.

STRATEGIC PATH

1. Prove Data Quality
   • Validate that gameplay-captured footage meets AI-lab QA standards (≥ 80 % pass rate).
   • Demonstrate verifiable labelling accuracy using hybrid AI + community review.
2. Prove Retention
   • Launch the “Daily Drop” loop + streak system.
   • Target 30-day retention > 20%, benchmarked above mobile gaming average.
3. Prove Virality
   • Activate influencer-driven rollout; hit 100 K MAU.
   • Validate share-rate > 20 % and organic invite chain growth.
4. Scale Supply → Demand
   • Build automated data marketplace and enterprise API.

CONSTRAINTS

• Platform must be profitable on data sales alone; consumer monetisation is secondary.
• Maintain ethical, privacy-safe collection (opt-in GPS, consent surfaces, anonymisation).
• Prioritise data quality over user volume until enterprise validation.
• Stay lean. No dependency on speculative VC-funded burn.

WHAT WE’RE NOT BUILDING

• ❌ A traditional mobile game chasing app-store charts.
• ❌ A micro-task or gig-economy platform.
• ❌ A short-form social network.

We’re a data-creation engine disguised as a global game.

ONE-SENTENCE PITCH

“We turn video players into the world’s largest distributed computer-vision dataset, one real-world challenge at a time.”

KEY PRINCIPLES

1. Data Quality > Scale – First six months focus on proof of precision, not vanity metrics.
2. Simple > Complex – Every feature must justify its psychological and data-value impact.
3. Async > Always-On – Encourage daily rituals, not endless scrolling.
4. Fun > Money – Intrinsic motivation builds better data and longer retention.
5. Transparent > Clever – Players should understand they’re helping train AI and be proud of it.

THE FLYWHEEL

1. Play: Daily challenges drop globally at fixed times.
2. Capture: Players record short clips matching prompts → instant AI pre-check.
3. Verify: Community votes confirm accuracy → data verified, users rewarded.
4. Reward: Points, streak multipliers, raffles, and social flex keep the loop hot.
5. Sell: Aggregated, labelled data licensed to AI labs → revenue → bigger prizes → more play.

THE PSYCHOLOGY ENGINE

• Streak & Multiplier System: Daily continuity amplifies points (Duolingo-style habit loop + loss aversion).
• Variable Reward Tiers: Common → Legendary challenges trigger dopamine-rich unpredictability while staying non-monetary.
• Social Combustion: Team raids, chain quests, and ghost-mode competition turn the world into a shared arena.
• Dynamic Prizes: Tiered sweepstakes, lightning rounds, and near-miss feedback keep short-term intensity.
• Flex Layer: Trophy cases, highlight reels, and regional dominance drive viral sharing and identity.

DATA AS THE PRODUCT

Every completed challenge yields:

• Timestamped, geo-tagged video with embedded metadata (lighting, device, context).
• Human verification data – a structured record of what humans agree a “cat”, “bus”, or “wave” is.

Each clip is a micro-label, together forming the world’s richest visual corpus.

MONETIZATION MODEL

1. Data licensing to AI labs, robotics firms, and vision researchers.
2. Sponsored Challenges for brand integration (“Film a Starbucks cup”).
3. Premium Pass – extra attempts, streak freezes, and gameplay perks.
4. Ad-supported free tier for accessibility.

Data sales remain the north-star revenue stream; gameplay monetisation sustains community.

WHY NOW

AI’s hunger for authentic visual data is exploding, while Gen Z’s gaming behaviour and influencer reach make scaled collection newly possible. No one sits at this intersection of data integrity × viral culture × mobile play.

By fusing the engagement mechanics of Pokémon GO and Duolingo with the data value of Scale AI, we can transform the economics of model training and let the world play a part in teaching intelligence.