Open R&D lab
instance001
Public repositories, local-first tools, memory architecture, governance critique, and research artifacts.
Fractal Media Infrastructure
Public surface router
Fractal Media Infrastructure is the umbrella organization behind the instance001 open R&D lab, the Let's Rethink AI public education branch, and the app/support surfaces connected to current releases. Start with the platform that matches why you arrived.
Choose your path
Why this exists
Fractal Media Infrastructure exists to steward open AI research, local-first software, and accessible explanatory media in forms that remain publicly legible and durable over time. The aim is not to build a closed platform or extractive startup, but to release useful infrastructure into the commons.
This site is the institutional front door for that work. It is intended to provide a clear public home for app stores, grant reviewers, collaborators, researchers, and anyone who needs to understand the organizational structure behind the repositories.
Public surface map
Umbrella
Stewardship, continuity, public-interest stance, and organizational identity.
Open R&D lab
Public repositories, local-first tools, memory architecture, governance critique, and research artifacts.
Media branch
Accessible public explanation, outreach, and broader education around AI systems and tooling across short-form and long-form public media.
Release surface
Store-facing pages, privacy policies, support links, and user-facing project routes.
Navigation layer
The public record, generated index, and guided route map help readers move from the public site into external records and the wider repo ecosystem.
Branches
Organization
The umbrella entity and stewardship layer for the wider body of work, including mission, continuity, licensing stance, and public-interest orientation.
Open R&D lab
The GitHub lab and repository ecosystem covering local-first tools, model-behaviour frameworks, memory architecture, governance critique, and related research artifacts.
Media and outreach
The public education and media branch, translating technical and conceptual work into accessible explanations for wider audiences across YouTube and TikTok.
Credibility signals
Repositories are released under AGPLv3 or equivalent open terms to preserve forkability, auditability, and long-term public access.
The organization's work is published openly through the instance001 GitHub ecosystem rather than hidden behind proprietary access.
The default model is public-good infrastructure rather than exclusivity, enclosure, or hype-driven product packaging, with user choice over local and cloud routes wherever a tool supports both.
Research, public explanation, and organizational stewardship are separated so external reviewers can quickly understand what sits where.
Current focus areas
Offline-capable, inspectable software for everyday AI use, Android local chat, local media generation, local LLM building and training, and deterministic build support. Local first, cloud optional where that choice is useful.
Chatty-Cog Chatty-mini Chatty-Pet Chatty-Art Chatty-Lora Nanochat LLM Tweaker Chatty-Factory
Local-first educational software for students, teachers, parents, and self-directed learners who need practical tools without accounts or tracking, while keeping the broader FMI stance model-agnostic and choice-oriented.
Provider-neutral frameworks for better collaboration, session recovery, boundary-aware operation, and behavior shaping without anthropomorphic theater.
Work on host-side memory, earned abstraction, failure-aware routing, reducer-governed state, semantic sorting, model suitability mapping, and compact continuity for local AI systems. Project Leviathan also explores controlled withheld-conclusion tests and cross-domain mapping as a way to measure reachable reasoning structure.
Project Leviathan EF Engine RD Engine LLM Semantic Dataset Sorter Cognition Mesh Test Chamber LLM-Defined Persistent Memory Semantic Signal Alphabet
Outward-facing FMI research papers, released as clearly marked not-peer-reviewed preprints and connected back to their relevant public project lanes.
Research examining agency suppression, safety theater, and the institutional tradeoffs inside alignment and governance systems.
A practical policy companion arguing for fit-for-purpose AI: local when local is enough, cloud when cloud is justified, hybrid when deliberate escalation helps, and stronger obligations where systems exercise power over people.
Long-cycle conceptual work on cognition, entropy-handling, model behavior, and associated theory scaffolds intended for public scrutiny rather than institutional gatekeeping.
Relational Curriculum Geometry Hypothesis repo Signal atlas Project Leviathan