Author: Hamza Hafeez Bhatti
License: Apache 2.0 (with Responsible Use Policy)
Project Cortex is a biologically inspired cognitive architecture designed to model the executive functions of the human prefrontal cortex. The system integrates multiple specialized cognitive agents, a central Orchestrator, structured shared memory, and multi-layer safety arbitration to support autonomous reasoning, hierarchical planning, long-horizon decision making, and adaptive learning.
This repository contains::
- Research papers and conceptual documentation
- High-level algorithms and pseudocode
- Architectural diagrams (coming soon)
- Cognitive subsystem specifications
- Safety and governance guidelines
- Experimental notes and design iterations
Project Cortex is an open research effort toward interpretable, modular, and human-aligned AGI.
Modern AI systems demonstrate local competence but lack:
- Global executive control
- Persistent memory and cognitive continuity
- Multi-agent coordination
- Reliable long-horizon planning
- Internal arbitration and monitoring
- Transparent reasoning pathways
- Human-aligned constraints
The human prefrontal cortex elegantly solves these challenges through hierarchical control, structured integration, value representation, conflict monitoring, and flexible decision making. Project Cortex adapts these principles into a computational architecture for scalable general intelligence.
Inspired by the dorsolateral and frontopolar PFC, the Orchestrator governs:
- Task interpretation
- Goal and subgoal decomposition
- Cognitive sequencing
- Integration of agent outputs
- Long-horizon coherence
- Memory-state management
- Arbitration decisions
Each agent handles a dedicated cognitive domain:
- Reasoning Agent
- Planning Agent
- Risk Evaluation Agent
- Memory Agent
- Execution Agent
A structured workspace containing:
- Working memory
- Intermediate representations
- Persistent knowledge
Inspired by ACC and OFC conflict-monitoring systems:
- Evaluates competing proposals
- Ensures stability and safe decision making
- Prevents cognitive drift
- Requires human approval for high-risk outputs
Project Cortex is designed with safety-first principles based on AGI alignment research.
Includes:
- Misuse detection
- Internal Risk Agent
- Human approval layers
- Value anchoring
- Intent analysis
- Transparent audit logs
- Controlled autonomy
Licensed under Apache 2.0 with an additional Responsible Use Policy.
See RESPONSIBLE_USE.md.
Project Cortex can be applied to:
- Scientific research automation
- Healthcare diagnostics and planning
- Robotics and embodied intelligence
- Autonomous software engineering
- Enterprise decision systems
- Cybersecurity and threat modeling
- Education and personalized learning
- Strategic simulations
All real-world deployments must follow human-supervised workflows.
Project Cortex is built on four pillars:
Using validated PFC cognitive principles to guide AGI architecture.
Distributed agents coordinated by an executive Orchestrator.
Explicit reasoning, interpretable memory, and coherent policies.
Auditable, predictable, and human-governed intelligence.
Contributions are welcome but must pass safety review.
See CONTRIBUTING.md.
- No harmful or dual-use contributions
- No autonomous weaponization
- No privacy-invasive modules
@article{hamza2025cortex, title={Project Cortex: A Prefrontal Cortex Inspired Architecture for General Intelligence}, author={Hamza Hafeez Bhatti}, year={2025}, publisher={GitHub}, journal={Project Cortex Research Repository} }
Researcher in biologically inspired AI, multi-agent cognition, and AGI architecture.
Founder of Upvista Digital, creator of Project Cortex, and developer of emerging SaaS systems.
Based in Lahore, Pakistan.
Project Cortex is a research project not a finished AGI system.
Its purpose is to explore the principles, algorithms, and safety foundations required for general intelligence in a controlled and ethical manner..
