v1.0.0 — Now Available

EPISTEME

AI Research Intelligence Layer

The extension that plugs into every scientific paper verifying claims, detecting bias, mapping knowledge graphs, and exposing what the authors didn't say.

Download Extension View on GitHub
episteme@research ~
analyzing paper: "Attention Is All You Need"
✓ CLAIM Transformer outperforms RNNs on WMT EN-DE by +2.0 BLEU
⚠ BIAS Self-citation rate: 34% — above threshold
✓ GAP Identified 3 open research directions
✓ GRAPH Concept map: 12 nodes, 18 edges rendered
ℹ PEER Mock review draft generated — 4 major concerns
_
15+
AI Features
arXiv
PubMed · Nature · IEEE
LangGraph
Pipeline Engine
NVIDIA
NIM LLaMA 3.1

Everything a researcher needs.
In one sidebar.

🔍

Claim Verification Engine

Each claim is extracted, scored for confidence, and cross-referenced with live data sources. Color-coded verdict badges instantly show what holds up.

🧠

Interactive Knowledge Graph

Real-time canvas-rendered force graph of concepts, methods, and relationships. Click any node to inspect connections and cluster semantics.

⚖️

Bias Meter & Integrity Score

Self-citation analysis, funding bias flags, methodology red-flags, and a statistical anomaly scanner — all surfaced automatically.

💡

Jargon Explainer

Highlight any term on the page. An AI-powered floating card explains it in plain language, with a one-click save to your Research Notebook.

🔭

Hypothesis Generator

Surfaces testable hypotheses from research gaps. One-click generates a full experimental protocol — complete with pseudocode and success metrics.

📓

Research Notebook

Save highlights, explanations, and notes across papers. Export everything as Obsidian-compatible Markdown — your second brain, built right in.

👤

Author Impact Network

H-index gauges, affiliation badges, co-author relationships, and top cited publications — all rendered in a scrollable author intelligence panel.

Evolution Timeline

Citation lineage visualized chronologically — see exactly how this paper's claims evolved from prior work, and what it contradicted or confirmed.

🤖

Chat with the Paper

Ask anything. The AI assistant has full paper context and gives grounded, precise answers — with honest "I can't find this in the text" admissions.

How the system works

01

Content Script Extracts

Injected into any research page. Extracts title, abstract, full text, authors, DOI, and arXiv ID automatically.

02

LangGraph Pipeline Fires

A 6-node multi-agent LangGraph DAG runs in parallel — claim extraction, trust scanning, hypothesis synthesis, replication finding, and more.

03

NIM LLaMA-3.1 Reasons

NVIDIA NIM serves LLaMA 3.1 70B and 8B at near-zero latency. Every synthesis, draft, and explanation is grounded and hallucination-resistant.

04

Sidebar Renders Intelligence

Results stream into the React sidebar — interactive graphs, scored claims, research gaps, and a personal notebook. Cached for instant re-opens.

Built with production-grade tools

⚛️
React + TypeScript
Extension UI
🐍
FastAPI
Backend API
🔗
LangGraph
Agent Orchestration
🟢
NVIDIA NIM
LLM Inference
🗄️
Qdrant
Vector Memory
Vite
Build Tooling
🤗
HF Spaces
Backend Hosting
🔺
Vercel
Website Hosting

Get Episteme
in 60 seconds

📦

Manual Install (ZIP)

Download the latest build and load it as an unpacked extension in Chrome or Edge.

1Download ZIP below
2Open chrome://extensions
3Enable Developer Mode
4Click Load Unpacked → select the unzipped folder
Download v1.0.0 ZIP
~267 KB · Chrome MV3 · No account needed

Build from Source

Clone the repo, install deps, and build locally. Full control over your environment.

bash
git clone https://github.com/Techie03/episteme
cd episteme/extension
npm install && npm run build
View on GitHub
Pre-configured Cloud Backend Ready. Episteme comes with a secure cloud backend out-of-the-box so you can start verifying papers immediately. For complete privacy or custom development, you can also run the FastAPI backend locally with uvicorn app.main:app or deploy it to your own Hugging Face Spaces and set the URL in the extension's Settings tab.