M1/M3
Reasoning machines for AI-and-memory substrates
What M1 and M3 Are
M1 and M3 are operational philomas for AI-and-memory substrates.
They are designed to help AI-supported work preserve structure across repeated sessions: reasoning, memory, uploaded materials, state, uncertainty, authority, recovery, and continuation.
The original public versions were written for ChatGPT Plus / ChatGPT Pro together with Google Drive. The structure can also be studied, rewritten, or adapted for other AI-and-memory substrates.
They are structured reasoning machines: each has a core, rules, state, commands, recovery procedures, and a way to continue work across interruptions - not prompts.
They are designed to help AI-supported work preserve structure across repeated sessions: reasoning, memory, uploaded materials, state, uncertainty, authority, recovery, and continuation.
The original public versions were written for ChatGPT Plus / ChatGPT Pro together with Google Drive. The structure can also be studied, rewritten, or adapted for other AI-and-memory substrates.
They are structured reasoning machines: each has a core, rules, state, commands, recovery procedures, and a way to continue work across interruptions - not prompts.
How They Relate to BOIS
BOIS provides the operating cycle — M1/M3 make that cycle operational on an AI-and-memory substrate.
BOIS defines the sequence of work: attention, gaps, questions, protocols, execution, experience, feedback, verified results, instructions, physiology, and renewed attention — M1/M3 translate that logic into working machine structures for AI-assisted reasoning, memory, uploaded documents, tool use, state recovery, and continuation.
In this sense, M1/M3 are working machine forms of BOIS. They carry the BOIS operating cycle into an AI-and-memory substrate, where reasoning, memory, authority, recovery, and continuation can be organized as structured work.
BOIS defines the sequence of work: attention, gaps, questions, protocols, execution, experience, feedback, verified results, instructions, physiology, and renewed attention — M1/M3 translate that logic into working machine structures for AI-assisted reasoning, memory, uploaded documents, tool use, state recovery, and continuation.
In this sense, M1/M3 are working machine forms of BOIS. They carry the BOIS operating cycle into an AI-and-memory substrate, where reasoning, memory, authority, recovery, and continuation can be organized as structured work.
What M1/M3 Preserve
M1/M3 are useful when AI work needs more than a single answer.
They are built for complex work that must preserve:
They are built for complex work that must preserve:
Reasoning
Structured reasoning with visible claims, grounds, authority, conclusions, and open questions.
Memory
Preserved working context, recovery rules, and continuation across sessions.
Materials
Work with provided files, sources, and tool-accessible materials while preserving provenance and scope.
State
A maintained record of the current task: accepted results, open questions, versions, positions, and obligations.
Recovery
Procedures for resuming after interruption, missing input, changed grounds, or error.
Continuation
A way for complex AI-supported work to keep structure across repeated tasks instead of restarting from a blank conversation.
M1 and M3: the Difference
- M1A single active line of reasoning.
M1 works as a sequential route processor. It keeps one active frame, one current position, and one accepted prefix of work. When a new inquiry is needed, the current line is suspended, a child inquiry is opened, and the result returns to the saved address before the main route continues.
This makes M1 useful when the main requirement is disciplined continuity: one question, one route, explicit returns, and a controlled path from input to accepted result. - M3A network of dependent reasoning cells.
M3 works through addressable cells and dependency-based readiness. Each cell has its own inputs, transformation, output, acceptance conditions, and dependencies. A cell becomes ready when its mandatory inputs of the required version have been accepted.
This makes M3 useful when the work contains several related branches. Independent cells can proceed when their dependencies allow it, while the final result is assembled only after the required dependencies of one version have been accepted.
How to Use M1/M3
Study them as examples of operational philomas
Use M1/M3 to see how a philoma can be organized as a working reasoning machine: with rules, state, memory, recovery, authority, and continuation.
Test them in an AI-and-memory environment
Run the materials in a suitable AI workspace with memory or stored files, and observe how the machine handles reasoning, uploaded materials, interruptions, open questions, and recovery.
Adapt
Use the public M1/M3 structure as a basis for rewriting the machine onto another AI-and-memory substrate. Adapt the implementation freely for your own system, while preserving the logic of substrate, memory, authority, recovery, and continuation.
The materials are public working releases, not a finished commercial product. They are provided for reading, testing, adaptation, and further development.
What the Public Materials Include
The public M1/M3 materials contain the working structure of the machines: their foundations, operating cycles, state and recovery logic, and configuration layers.
- FoundationsStartup and execution boundaries, theoretical basis, terminology, organs, levels, and the basic organization of the philoma.
- Operating StructureThe operating cycle of the philomic approach.
- State and RecoveryState, contact, debt, recovery, repair, completion, memory, and the Execute Sleep command.
- Testing and ConfigurationExamples, self-test, active rule registry, two-layer core loading, and personal layer configuration.
Public Release
Download the free public M1/M3 release or view the repository.
The release contains the public materials for M1/M3 as reasoning machines for AI-and-memory substrates. The package includes a README file in both Russian and English with practical instructions for starting work with M1/M3.
The release contains the public materials for M1/M3 as reasoning machines for AI-and-memory substrates. The package includes a README file in both Russian and English with practical instructions for starting work with M1/M3.
Related Materials
- God and the auditorThe book-length synthesis of the research and BOIS design work.
- BOIS
The operating cycle of the philomic approach. - Pravda
The educational philoma presented in the book. - MaOS
A separate playable project that demonstrates BOIS through play.
