Paste an AI answer and see which claims are supported by real sources
Advanced26h2–3
LLMSearch APIRAG
16/20Docs to Knowledge Graph Explorer
Turn a folder of documents into a clickable graph of people, places and claims.
Advanced28h2–4 people15/20 fit
Build Docs to Knowledge Graph Explorer for researchers and analysts in ~28h with 2–4 people. Stack: LLM + Python + ECharts + SQLite.
Why it wins: Graph visuals look great and every edge links to evidence.
Build flow
01 · Problem
Connections between entities across many documents are invisible in plain search.
02 · Solution
Extract entities and relations with an LLM, merge duplicates, store them in a graph and visualise it with source snippets on every edge.
03 · Build
- 1Chunk documents
- 2Extract triples with the LLM
- 3Merge entities by name and embedding
- 4Render with an ECharts graph
04 · Demo
Load 30 news articles and click a person to see every linked organisation with sources.
MVP checklist
- Entity and relation extraction
- Duplicate merging
- Interactive graph
- Source snippets per edge
Tech stack
| Tech | Role | |
|---|---|---|
| LLM | Language model | AI |
| Python | Backend / scripting | Backend |
| ECharts | Interactive charts | UI |
| SQLite | Embedded database | Data |
Reference repos
Open-source projects you can build on. Links checked 2026-10-31.
Related ideas
More AIAI Agents
Multi-Agent Research Brief
Produces a one-page cited brief on any question in minutes
Advanced36h2–4PythonLangGraphSearch API16/20Jumps to the exact second in a long video where something was said or shown
Advanced48h2–4PythonWhisperVector DB16/20