MyAIAssistant¶

Project Goals¶

MyAIAssistant helps users organize tasks, reference subject-matter knowledge, and leverage AI for semantic search, note summarization, task extraction and recommendations. The tool links knowledge artifacts to tasks to provide better context when addressing work.

Based on Stephen Covey's "7 Habits of Highly Effective People" the system helps manage priorities efficiently using the Eisenhower Matrix (Urgent/Important classification).

Project Principles¶

1. Run locally - All core features work without external API dependencies
2. External configuration - Environment-based configuration for flexibility
3. Privacy-first - Data stays on local infrastructure
4. Efficient prioritization - Eisenhower Matrix helps focus on high-impact work

With a drag-and-drop user interface it is easy to continuously re-prioritize tasks. With AI assistant, the tool helps to better run a working day, query what was done last wee, last month, last quarter...

The goal is to run locally, with local or remote LLM, and using isolate environment or workspace to separate works, from learning or home work.

Access the webApp local once started.

Why I built this?¶

There are ton of to do apps, why a new one? The main drivers for creating this app:

• How I can augment myself and my day-to-day outcomes?
• Address what are the things I can do more, related ot what I value? I think the answer is learning, but how to organize time for it, and how to really track, learning activities?
• I used paper or digital notes based on the Eisenhower matrix to track my tasks, but I think, AI can enhance the plannification, the decomposition and the semantic search of my current knowledge, my past work, and projects.
• Address how to maximize my time on desired activities and minimize time on overhead?
• While tracking task with app, it is easy to build weekly report, an essential tool for management.
• NotebookLM, Glean, RAG, GraphRAG... are excellent tools to start managing knowledge and query it, but they are not related to my work directly. I think it may be possible to fill this gap, by linking all the information about organizations, projects, tasks, assets all together.
• The tool should also helps to address the efficient use of AI, by helping the step of clear thinking (getting a clear statement of the problem), clear writing (by formalizing in simple sentences the problem, to explain what you want to do), to build prompts for more efficient AI.
• Apply agentic architecture for a simple day-to-day application, as a learning experience.

Core Features¶

Quick Start¶

See user guide

Understanding the implementation¶

See design and implementation chapter.

Activities¶

Create Knowledge base¶

Tool to vectorize folder content into ChromaDB for RAG.

This script scans a folder for supported documents (markdown, text, HTML), processes them through the RAG pipeline, stores embeddings in ChromaDB, and saves metadata to the knowledge base database.

The tool is reentrant: re-running it will update existing documents if their content has changed, and skip unchanged files.

Features¶

Code is: backend/tools/vectorize_folder.py

This CLI tool leverages the existing backend services:

• DocumentLoader for loading markdown and HTML files
• RecursiveTextSplitter for chunking content

• ChromaDB with all-MiniLM-L6-v2 embeddings

• Recursive folder scanning for supported file types

• Configurable chunk size and overlap
• Category and tags metadata support
• Progress logging with file-by-file status
• Collection statistics view
• Handles .md, .markdown, .txt, and .html files