4mind
4mind turns uploaded documents into structured insights.
Core trick
Gemini extracts text into a knowledge graph, computes embeddings, infers ontology edges, and fills a python-docx template.
Description
4mind turns uploaded documents into structured insights. The FastAPI backend extracts text, builds knowledge graphs, computes embeddings, infers ontology edges, summarizes relationships, and now assembles a science-style article with a DOCX download. Gemini drives LLM steps; Python docx fills a template. The Next.js frontend (static export) lets users upload files, stream graph/embedding/inference progress, review summaries, and trigger article generation, all with shareable visualizations.
README excerpt
# 4mind See live demo: https://4mind.tech/ ## Plan on the day 1. Front-end - [x] "Wow" animation on the start (something like https://www.vyomaui.design/backgrounds/circuit-flow) - [x] Upload file(s) with a progressbar - [x] List files (and remove) - [x] Start generating (pooling status / delta) - [x] Display delta (graph chart, embeddings chart) - [x] Start extract new ideas - [x] Visualize new insights 3. Back-end - [x] User/projects end-points (create/list) - [x] Files end-points (upload/remove/download) - [x] Graph end-points (start/status-delta/list of all nodes) - [x] Embeddings end-points (start/status-delta/list of all nodes) - [x] Insights (`sync_reasoner()`) end-points (create/status-delta/list) - [ ] ~~Export charts / data / report~~ ### Before 11:00 AM - [x] Set up project repository - [x] Prepare the domain name purchase ### Before 3:00 PM - [ ] Dev ### Before 4:00 PM - [ ] Test - [ ] Improve UI/UX ### Before 5:00 PM - [ ] Record demo video - [ ] Upload video to YouTube - [ ] Prepare description - [ ] Prepare speech notes