Building an Artificial Cognitive Architecture from Scratch.
BitGenesis is an open-source research project focused on designing and implementing an artificial cognitive architecture entirely from scratch.
Unlike traditional AI assistants that rely on pre-trained Large Language Models (LLMs), BitGenesis aims to develop its own modular cognitive system, where every component is designed, implemented, tested and documented independently.
The long-term vision is to create an intelligent software architecture capable of perceiving information, storing knowledge, reasoning, planning actions, learning from experience and interacting with external tools through a transparent and explainable design.
The primary objectives of BitGenesis are:
- Build an artificial cognitive architecture from scratch.
- Design every subsystem as an independent and replaceable module.
- Develop explainable reasoning instead of opaque decision making.
- Implement flexible short-term and long-term memory systems.
- Create an event-driven architecture.
- Allow continuous learning and knowledge acquisition.
- Integrate external development tools without depending on proprietary AI services.
- Maintain a fully documented and reproducible codebase.
BitGenesis is developed around the following principles:
- Modularity – Every subsystem has a single responsibility.
- Transparency – Every decision should be explainable.
- Extensibility – Components can evolve independently.
- Determinism – Predictable behavior whenever possible.
- Documentation First – Architecture is designed before implementation.
- Security – Safe interaction with external systems.
- Maintainability – Long-term sustainability over short-term complexity.
Current version:
v0.1.0
Development stage:
Foundation Release
BitGenesis has reached its first functional milestone.
The current implementation includes the foundations of a modular cognitive architecture:
- Cognitive core and Brain lifecycle
- Event-driven architecture
- Identity system
- Memory system
- Memory retrieval and importance evaluation
- Memory consolidation and episode generation
- Knowledge representation
- Reasoning engine
- Reflection system
- Inference engine
- Dialogue and response system
The architecture is continuously evolving toward a more complete cognitive framework.
- Brain controller
- Lifecycle management
- Configuration system
- Runtime statistics
- Version management
- Event-based memory creation
- Memory storage
- Memory querying
- Memory retrieval
- Memory similarity
- Importance scoring
- Memory consolidation
- Episode generation
- Entity management
- Knowledge graph foundation
- Relations
- Knowledge queries
- Inference rules
- Intent detection
- Resolution system
- Reflection engine
- Inference engine
- Reasoning sessions
- Identity responses
- Memory self-report
- Response formatting
- Dialogue engine
Clone the repository:
git clone https://github.com/your-user/BitGenesis.git
cd BitGenesisInstall the package:
pip install -e .from bitgenesis.core.brain import Brain
brain = Brain()
response = brain.ask(
"What do you remember?"
)
print(response)BitGenesis currently includes:
263 automated tests passing
Run the test suite:
pytestbitgenesis/
├── core/
├── cognition/
├── memory/
├── knowledge/
├── reasoning/
├── planning/
├── dialogue/
├── events/
├── runtime/
├── kernel/
├── learning/
├── language/
├── perception/
├── neural/
├── tools/
├── security/
└── utils/
Additional project documentation can be found inside the docs/ directory.
BitGenesis follows an architecture-first approach.
Every major component is:
- Specified
- Documented
- Implemented
- Tested
- Reviewed
before becoming part of the official architecture.
Completed:
- Core architecture
- Memory subsystem
- Knowledge subsystem
- Reasoning foundation
- Reflection and inference
- Dialogue system
- Testing infrastructure
Planned:
- Kernel-driven architecture
- Persistent memory storage
- Improved cognitive pipeline
- Autonomous lifecycle management
- Learning subsystem expansion
- Tool execution framework
Future versions will expand toward:
- Advanced planning
- Language understanding
- Autonomous agents
- External environment interaction
Contributions are welcome.
Please read the CONTRIBUTING.md document before submitting issues or pull requests.
BitGenesis is licensed under the Apache License 2.0.
See the LICENSE file for details.
BitGenesis is not intended to be another chatbot.
Its purpose is to explore how an artificial cognitive architecture can be designed from first principles, combining software engineering, reasoning systems, memory structures and learning algorithms into a transparent, modular and extensible platform.
