Skip to content

calin-tesu/Momentum

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

39 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Momentum

Momentum is an agentic learning companion designed to help self-directed developers resume progress when motivation drops or learning is postponed.

The system is intentionally simple and bounded.


Core Idea

Momentum does not generate curriculum, plans, or long-term goals.

Instead, it:

  • operates on a fixed learning path defined by the developer
  • uses deterministic rules to detect inactivity or repeated postponement
  • delegates the choice of the next small action to an AI agent

The purpose of the AI is to make continuation easier than quitting.


Role of AI (Important)

The AI agent:

  • does not decide what to learn
  • does not generate new learning steps
  • does not optimize for speed or completion

The AI only:

  • selects one task from a bounded task template library
  • adapts its scope based on user behavior
  • phrases the task in a low-pressure, human-friendly way

All curriculum structure, rules, and constraints are deterministic.


System Overview

  1. The user follows a predefined learning path (e.g., Android / Jetpack Compose)
  2. The system tracks completion, postponement, and inactivity
  3. Rule-based logic selects an intervention strategy
  4. The AI agent selects and instantiates one minimal next action
  5. The user either completes or postpones the action

Only one task is presented at any time.


Design Principles

  • Small steps beat perfect plans
  • Restarting matters more than speed
  • Scope reduction is preferable to pressure
  • AI judgment is bounded and explainable

Explicit Non-Goals

Momentum does not aim to:

  • replace structured courses
  • act as a general productivity tool
  • monitor external activity (e.g., GitHub)
  • provide motivational coaching or reminders

Evaluation

The project is evaluated using Comet’s Opik, focusing on:

  • appropriateness of selected tasks
  • quality of scope adaptation
  • alignment between strategy and action

Completion rates are not the primary success metric.


This README is intentionally concise to serve as ground truth for both developers and AI-assisted tools.

About

An agentic learning companion that helps developers resume progress after procrastination by selecting a single, minimal next action from a bounded task space. Focused on re-entry, not productivity, with deterministic rules and explainable AI decisions evaluated via Comet Opik.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages