Featured Project
Shard One Master Suite
The Shard One Master Suite bundles all of our flagship tools—including NeuroFlow, MindSpell, Cognitive State Tracker, and more—into a single, easy-to-install package. Designed for researchers, developers, and enthusiasts, the suite provides seamless integration between BCI, neurofeedback, cognitive analytics, and agentic AI applications.
Free Tier: Use any combination of local models you want—enjoy full access to the suite's core features with your own on-device AI, at no cost.
Premium Option: A subscription unlocks advanced agentic capabilities powered by Claude or GPT, enabling next-level research and automation within our framework.
Nexus Includes:
- NeuroFlow: Real-time BCI/EEG platform
- MindSpell: BCI spelling and communication
- Cognitive State Tracker: Cognitive analytics
- Spatial Reasoning Agent: Embodied AI navigation
- Multi-Agent Reasoning System: Collaborative AI
- And more—see below for details
NeuroFlow
A comprehensive BCI platform that connects Open Neural Interface or OpenBCI devices with real-time ML processing. Features custom neural network architectures for EEG signal classification, mental state detection, and neurofeedback applications. Built with TensorFlow and optimized for real-time performance.
Key Features:
- Real-time EEG processing with custom CNN architectures
- Mental state classification (focus, relaxation, meditation)
- Adaptive neurofeedback protocols
- Open Neural Interface Cyton/Ganglion integration
- Pre-trained models for common tasks
- Training pipeline for custom datasets
Spatial Reasoning Agent
An AI agent that navigates and understands 3D environments using spatial cognition principles. Can be deployed in Unity/Unreal simulations or connected to robotic platforms. Features memory-based spatial mapping and goal-directed navigation.
Features:
- 3D spatial mapping and memory formation
- Goal-directed navigation with obstacle avoidance
- Unity/Unreal Engine integration
- ROS compatibility for real robots
MindSpell
A practical BCI spelling application that uses motor imagery and P300 potentials to enable hands-free text input. Features adaptive ML models that learn from user patterns and improve accuracy over time.
Features:
- P300 and motor imagery classification
- Adaptive learning from user sessions
- Word prediction and autocomplete
- Customizable interface layouts
Embodied Interaction Lab
A virtual laboratory for training embodied AI agents that understand gesture, spatial relationships, and multimodal communication. Agents learn to interact naturally through embodied experience rather than pure text.
Features:
- Gesture recognition and generation
- Spatial context understanding
- Multimodal interaction training
- Embodied language grounding
Autonomous Observer
A sensory-focused AI system that autonomously observes and analyzes research environments. Equipped with vision, audio, and environmental sensors to collect and interpret data from ongoing experiments without direct intervention capabilities.
Features:
- Multi-modal sensory data collection
- Real-time environmental monitoring
- Behavioral pattern recognition
- Autonomous observation scheduling
Cognitive State Tracker
A comprehensive system that combines EEG data with behavioral metrics to track cognitive states throughout the day. Uses ensemble ML models to predict focus, fatigue, and cognitive load from multimodal inputs.
Features:
- Multi-modal data fusion (EEG + behavioral)
- Real-time cognitive state estimation
- Productivity optimization suggestions
- Long-term pattern analysis
Multi-Agent Reasoning System
A hybrid architecture platform where specialized local models handle data collection, organization, and preprocessing, while large language models provide high-level reasoning and decision-making. Agents collaborate through structured data pipelines and semantic communication protocols.
Features:
- Local models for data collection and organization
- LLM integration for strategic decision-making
- Hybrid local-cloud reasoning architecture
- Semantic inter-agent communication protocols
Attention Optimizer
A closed-loop neurofeedback system that monitors attention levels during work and provides real-time interventions. Uses reinforcement learning to optimize feedback protocols for individual users.
Features:
- Real-time attention monitoring
- Adaptive feedback protocols
- Integration with productivity apps
- Personalized intervention strategies
Open Castle
An open-source AI-powered anti-cheat system for online games. Uses behavioral analysis and machine learning to detect cheating patterns in real time, supporting integration with popular game engines and server platforms.
Features:
- Real-time behavioral cheat detection
- Integration with Unity, Unreal, and custom engines
- Adaptive learning from new cheat methods
- Open-source and community extensible
Laera Eris
An experimental massively multiplayer online game (MMO) driven by advanced AI agents. Features emergent world-building, dynamic NPCs, and player-AI collaboration in a persistent, evolving universe.
Features:
- AI-driven NPCs with evolving personalities
- Procedural world generation and events
- Collaborative quests and storylines
- Player and AI co-creation of game content