NotesBuddy: AI-Powered Personal Note Search Tool
A developer built NotesBuddy, a private search and answer tool using Gemma 3, to help a friend organize scattered study notes effortlessly.
Stock photo for illustration only, not from the actual event
- NotesBuddy was created as a submission for the Hacktoberfest Weekend Challenge.
- It aggregates scattered text files and answers questions using plain language.
- Powered locally by Gemma 3 via Ollama to generate accurate responses.
- The entire project consists of a single Python file with zero external dependencies.
Managing study notes scattered across numerous files can be an exhausting task. Alekh, a friend of the developer, faced this exact dilemma where locating a small concept took an unreasonable amount of time. To solve this, a dedicated study assistant tool was built to streamline information retrieval.
NotesBuddy operates through a straightforward workflow. Users simply drop their notes into a designated folder and ask questions in natural language. The system retrieves a concise answer alongside the exact passage from the notes as a citation. Crucially, if the information is missing from the provided files, the tool explicitly states that it does not know rather than fabricating an answer, preventing misleading output.
Stock photo for illustration only, not from the actual event
From an architectural standpoint, the entire project resides in a single Python file requiring no dependencies outside the standard library. A retrieval-only live demo is hosted on Render to showcase search functionality and citations, though the full version utilizing Gemma for answer generation runs locally due to hosting limitations.
"Honestly, NotesBuddy is exactly what I needed. I've got so many notes scattered across different files that finding one small concept takes forever. Now I can just ask a question and get the answer along with the exact part of my notes it came from."
Alekh
Developing lightweight Retrieval-Augmented Generation (RAG) tools like this highlights the power of personal knowledge management combined with localized open-weights AI. Enforcing strict grounding—where the model refuses to answer if context is absent—is a vital design choice that addresses LLM hallucination in practical scenarios.
Looking ahead, the creator plans to swap BM25 search for local embeddings to better handle paraphrased queries, alongside introducing a quiz mode that transforms raw notes into interactive practice questions for revision.
Source: Dev.to
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment