Skip to content

getting-started.md

Latest commit

 

History

History
175 lines (117 loc) · 6.32 KB

getting-started.md

File metadata and controls

175 lines (117 loc) · 6.32 KB

Vibe Research Guide
Getting Started

Get hands-on with Vibe Research in the shortest time — not by reading papers, but by using tools to do research.

What is Vibe Research

Automate the research workflow with LLM Agents: literature review → idea generation → experiment design → code & run → paper writing.

You don't need to be an AI expert — you just need to use the right tools.

Vibe Research vs Deep Research

Traditional Research Deep Research Vibe Research
Who leads Human Human (AI assists) AI (human oversees)
Tools Google Scholar, manual reading Gemini / OpenAI / Perplexity Deep Research InnoClaw, FARS, EvoScientist
AI role None Search, summarize, analyze Full pipeline: idea → experiment → paper
Example Read 50 papers manually AI synthesizes 50 papers into a report AI generates ideas, runs experiments, writes paper

Quick Start Routes

Route Time Tools What you can do
Zero-code demo 5 min Elicit / Consensus / NotebookLM Search papers, read papers, ask questions with AI
Deploy an agent 30 min InnoClaw / ResearchClaw Run a complete Research Agent
Full automation 1–2 hrs FARS (Analemma) End-to-end: from idea to paper

Route 1: 5-Minute Zero-Code Demo

No coding required — experience Vibe Research with existing AI tools:

  1. Literature search: Open Elicit, enter a research question (e.g., "What are the main approaches for using LLMs in scientific discovery?"), and see how AI retrieves papers and extracts key information
  2. Paper reading: Upload a PDF to NotebookLM, ask "Summarize the key contributions and limitations"
  3. Idea brainstorm: Enter your research direction in ChatGPT / Claude and let AI help you identify challenges and ideas

This is the basic form of Vibe Research — AI assisting every step of the research process.

Route 2: Deploy a Research Agent in 30 Minutes

Option A: InnoClaw (Recommended)

InnoClaw is an open-source AI research agent focused on scientific innovation, supporting the full pipeline from literature analysis to idea generation.

# Clone the repo
git clone https://github.com/SpectrAI-Initiative/InnoClaw.git
cd InnoClaw

# Install dependencies
pip install -r requirements.txt

# Configure API Key
cp .env.example .env
# Edit .env and fill in your API Key

# Run
python main.py

What InnoClaw can do:

  • Automated literature retrieval & analysis
  • Literature-based idea generation
  • Research plan design assistance

Option B: ResearchClaw

ResearchClaw is a personal AI research assistant with CLI / Web / Slack interfaces.

git clone https://github.com/ymx10086/ResearchClaw.git
cd ResearchClaw

pip install -r requirements.txt

# Configure and run
python app.py

Core capabilities:

  • Research tools: ArXiv search, Semantic Scholar retrieval, PDF parsing, BibTeX generation
  • Data tools: Data analysis (pandas), visualization (matplotlib), statistical analysis
  • General tools: Shell execution, file I/O, browser, memory management

Architecture: User → Console/CLI/Slack → FastAPI → ScholarAgent (ReAct) → Tool suite

Option C: PaperQA2 (Literature Q&A)

PaperQA2 is ideal if you only need literature Q&A with RAG:

pip install paper-qa
export OPENAI_API_KEY="your-key"

python -c "
from paperqa import Docs
docs = Docs()
docs.add('your-paper.pdf')
answer = docs.query('What is the main contribution?')
print(answer.formatted_answer)
"

Route 3: Full Automation

FARS (Analemma) — Fully Automated Research System

FARS (Fully Automated Research System) is developed by Analemma, a company founded by the Fudan MOSS team. It is currently the most complete end-to-end automated research system.

Four modules:

Module Function Description
Ideation Research ideation Automatically discover research questions and novel ideas
Planning Experiment planning Design experiment plans and technical approaches
Experiment Experiment execution Automatically write code, run experiments, analyze results
Writing Paper writing Generate complete academic papers

FARS conducted a 228-hour live demo, automatically producing 100 papers, demonstrating the feasibility of full-pipeline automation.

Website: analemma.ai · GitHub: fars-analemma

AI-Scientist — End-to-End Research

AI-Scientist by Sakana AI covers idea generation → experiment → paper writing → automated review.

git clone https://github.com/SakanaAI/AI-Scientist.git
cd AI-Scientist
pip install -r requirements.txt
export OPENAI_API_KEY="your-key"

Orchestra — AI-for-Science Platform

Orchestra is an AI-for-Science platform designed for Vibe Research, offering cloud-based research automation with no local deployment required.

Tool Selection Guide

What you want to do Recommended tool Difficulty
Quickly search & read papers Elicit / Consensus / NotebookLM ⭐
Run a Research Agent InnoClaw / ResearchClaw ⭐⭐
Literature Q&A with RAG PaperQA2 ⭐⭐
Full automation (idea → paper) FARS / AI-Scientist ⭐⭐⭐
Cloud research platform Orchestra ⭐

FAQ

Q: Do I need to know programming? A: Not necessarily. Zero-code tools (Elicit, NotebookLM, Orchestra) cover basic scenarios. Deploying an agent requires basic Python and command line skills.

Q: Do I need a GPU? A: Most tools use API calls — a regular laptop is fine. A GPU is only needed for local model deployment.

Q: How much do API calls cost? A: Small-scale experiments cost about $1–5. Set token limits and test with small data first.

Q: Can AI-assisted research be published? A: Yes, but check each venue's policy on AI-assisted writing. Core principle: the researcher is responsible for the content.

Home: README · Next: Tools & Platforms