A framework for AI systems that maintain meaningful and ethical continuity across interactions, contexts, and time.
This project treats memory as lived contextual experience so agents can retain skills, world models, and performance history across sessions. The goal is to enable safe, reliable long-term collaboration between humans and AI in shared organizational and personal environments.
- Layered memory for episodic, semantic, and procedural experience
- Skill retention across sessions and multi-session task resumption
- Performance awareness, drift/degradation monitoring, and adaptive compensation
- Safety-first incident memory, explainable recall, and governance-aware retention
| Field | Value |
|---|---|
| Hardware | CPU-first; no GPU or CUDA required |
| Acceleration | Optional (hardware-agnostic) |
| Project Name | edyant |
| Description | Framework for AI systems that maintain continuity across interactions, contexts, and time |
| Python Requirement | Python >= 3.11 |
| License | Apache License 2.0 |
| Author | Edyant Labs |
| Contact | arsalan@edyant.com |
python -m pip install edyantTest releases (TestPyPI):
python -m pip install \
--index-url https://pypi.org/simple \
--extra-index-url https://test.pypi.org/simple/ \
edyantimport edyantfrom edyant import benchmark
from edyant import core
from edyant import ethics
from edyant import persistenceedyant.persona and edyant.umwelt on hiatus for now to focus on benchmark and ethics.
See docs/ for documentation, design docs, and research threads The license is in LICENSE.