Play. Earn. Train. Computer-Use Agents.
Turn powerful foundation models into capable computer-use agents.
Open-source has ZERO desktop computer-use datasets. We're changing that with the Training Gym - the ONLY platform that collects high-quality data for fine-tuning state-of-the-art models like DeepSeek and Qwen VL into powerful computer-use agents. Get instantly rewarded for contributing to the future of AI.
Collective Intelligence Platform
Training powerful agents is now efficient and rewarding, powered by the $VIRAL token ecosystem
Play Tasks
Jump into our Training Gym and complete AI-generated computer tasks. From games to apps, every click and action contributes to something bigger.
Earn Instantly
Your gameplay automatically becomes high-quality training data. We augment UI with reasoning, and you get $VIRAL tokens instantly for your contributions.
Create Agents
Turn your data into powerful computer-use agents that understand any UI. One-click trains DeepSeek into your own LAM that can reason, navigate, and execute tasks just like you do.
Meet VM-1
Coming soon: Your AI companion that controls the full desktop - locally or in the cloud. Run on anything from games to productivity apps, and continually learning skills from the Training Gym. Join the telegram to be first in line.
Ready to help! What are we playing today? 🎮
Join my CS:GO match 192.168.1.1:27015
Play like Gold Nova 1 so I can practice!
Let's crush it! 🎯
Don't Believe Us?
Train your own LAM in minutes. Keep the weights.
Success rate on office UI tasks vs 0% with traditional OCR - tested on real computer-use benchmarks
Tasks to train a capable agent through our Training Gym
Deploy your trained agent to any task
Build Your Own LAM
- 1 Play games & complete tasks in our Training Gym
- 2 Your gameplay becomes high-quality training data
- 3 One-click fine-tuning creates your custom LAM
- 4 Deploy & earn $VIRAL as others use your agent
$VIRAL Ecosystem
HW7D5MyYG4Dz2C98axfjVBeLWpsEnofrqy6ZUwqwpump Building the future of AI agents through collective intelligence
Data Economy
- Fair compensation for high-quality training data - users earn $VIRAL for contributing validated behavioral demonstrations
- Cost-effective data collection through token incentives reduces training costs compared to traditional methods
- Transparent data marketplace where contributors maintain ownership and earn from dataset usage
Frontier Tech Development
- Building next-gen LAMs that understand any UI, game, or application, essential for business automation
- Advanced research into behavioral learning & real-time action prediction
- Revolutionary AI companion tech - your agent plays games with you in real-time
Technical Architecture
- Behavioral cloning with automated validation ensures high-quality demonstrations
- One-click deployment handled by our infrastructure - we manage the complexity so you don't have to
- Native integration with all major agent frameworks & crypto platforms
Community-Driven Growth
- Massive dataset expansion through gamified Training Gym
- Strategic partnerships with gaming studios & crypto projects
- Accelerating AI agent ecosystem through $VIRAL incentives
The Future is Here
Join us in building the next generation of AI agents - capable of using any computer interface just like humans do.
Who We Are
We're a team of passionate AI researchers and engineers from MIT, Microsoft, and leading AI labs, united by a vision to democratize frontier AI research.
Our mission is to bring powerful Large Action Models to the crypto community, creating a decentralized ecosystem where anyone can train and deploy truly capable AI agents.