Deterministic learning systems for writing and reasoning.

The Digital Learning Companion (DLC) is an open-source structured learning system built on the Emergent State Machine (ESM) architecture.

It demonstrates how instructional AI can be:

• deterministic
• auditable
• policy-governed
• replayable
• transparent to teachers and students

Rather than relying on opaque LLM behavior, the DLC models learning situations explicitly and applies structured instructional policies.

➡ DLC Meta Repository

https://github.com/Digital-Learning-Companion/dlc

This repository launches the full system and contains the Bird + Brain architecture.

The DLC is a reference implementation of the following architecture stack.

Control architecture for situational reasoning systems.

https://github.com/emergent-state-machine
https://github.com/emergent-state-machine/esm-spec

Framework for policy mutation in deterministic systems.

https://github.com/controlled-mutation-layer

Educational reference implementation demonstrating deterministic instructional guidance.

https://github.com/Digital-Learning-Companion/dlc

The DLC shows how AI learning systems can implement a full structured reasoning loop:

Observation
→ Signal extraction
→ State construction
→ Instructional gate
→ Policy projection
→ Instructional action
→ Learning evolution

This allows learning systems to provide guidance that is:

• interpretable
• stable
• inspectable
• versioned

instead of drifting through hidden model behavior.

The current DLC MVP implements a structured writing tutor that supports the development of:

Claim
Evidence
Reasoning

in argumentative writing.

The system analyzes writing structure, tracks learning signals, and delivers deterministic instructional guidance while recording interpretable telemetry.

This project is intended for:

• learning scientists
• education technologists
• engineers building structured AI systems
• organizations requiring auditability in AI-assisted workflows

Active development.

The writing tutor MVP is operational and the control architecture is stable.