Skip to content
UXClaim
User Research

Research Team Tutorial

Build a multi-agent AI research system using Claude Code that extracts data from papers, performs analysis, writes articles, and reviews its own work.

What It Does

This tutorial teaches you to create a coordinated team of AI agents that work together on research tasks. Your agents will extract data from academic PDFs, calculate statistics, write professional articles with proper citations, and review each other’s work for quality.

How It Works

The system uses Claude Code to create specialized agents:

  • Researcher Agent: Extracts data from PDFs and performs statistical analysis (correlations, regressions, effect sizes)
  • Copywriter Agent: Writes professional articles with proper academic formatting
  • Antagonist Agent: Reviews outputs and provides critical feedback
  • Orchestrator Agent: Coordinates the workflow between specialized agents

Agents communicate through skill definitions and use Python for data processing. The Model Context Protocol (MCP) securely accesses your research files.

Use Cases

  • Analyzing relationships across multiple research papers
  • Creating reproducible research workflows
  • Training teams on agentic AI concepts
  • Building quality control loops into automated systems
  • Coordinating complex multi-step research projects

Who Benefits

UX researchers and design practitioners benefit by learning to orchestrate AI agents for research synthesis, automated analysis, and report generation. Product managers can use this pattern for user research workflows. No coding experience needed—the tutorial guides beginners through every step.

Frequently asked questions

What do I need to install?
You need Claude Code 2.0+, Python 3.11+, and basic terminal knowledge. The tutorial provides step-by-step installation instructions for dependencies like PyPDF2, scipy, and numpy through requirements.txt.
How long does the tutorial take?
The complete workflow takes 10-15 minutes to run once set up. Initial setup (installing dependencies, configuring MCP server) takes 5-10 minutes. The tutorial guides include troubleshooting for common issues.
What research papers do I need?
You'll need three academic papers on cybersecurity and professional fatigue: Stanton et al. (2016), Reeves et al. (2021), and Mizrak et al. (2025). The tutorial explains where to find them and how to download them for your analysis.
Can I use this for my own research?
Yes. The system is designed to be adapted for different research tasks. You can modify the skill definitions to analyze your own papers and customize the analysis and article-writing requirements.
What is a multi-agent system?
A multi-agent system uses multiple specialized AI agents that work together on complex tasks. Each agent has specific expertise, they coordinate through the orchestrator, and they can review and improve each other's work—improving quality and handling tasks that require different types of knowledge.
Do I need to know statistics?
Helpful but not required. The tutorial handles statistical calculations automatically using Python. You'll understand the concepts (correlation, p-values) but don't need to write the code yourself.
What's the Model Context Protocol (MCP)?
MCP is a secure way for Claude Code to access resources like files and tools on your computer. In this tutorial, it lets the AI agents read your research papers and run Python scripts for analysis.
How do agents coordinate their work?
The orchestrator agent delegates tasks to specialists using Claude Code's Skill tool. Agents can pass data between each other, and the antagonist provides feedback that triggers revision loops until quality standards are met.

Glossary

Skill
An AI agent definition in Claude Code that gives an agent specific capabilities and expertise. Skills are defined in YAML and let agents specialize in particular tasks like data extraction or article writing.
Orchestrator
The lead agent that coordinates workflow between specialized agents. It delegates tasks, manages data flow, and ensures the overall process runs smoothly from start to finish.
Model Context Protocol (MCP)
A secure framework that lets Claude Code safely access external resources like files, databases, and tools. In this tutorial, MCP allows agents to read PDF files from your computer.
Agent
A specialized AI instance with particular expertise and capabilities. Agents in this system have distinct roles (researcher, copywriter, reviewer) and work together to complete complex research tasks.
Revision Loop
A quality control process where the antagonist agent reviews work and provides feedback, triggering the original agent to revise until quality standards are met or approval is granted.

More in User Research

All →
User Research

Ditto Product Research

Run customer research, pricing tests, and product validation using 300K+ AI personas with 92% focus group overlap.

Ask-Ditto
User Research

Reddit Intel

Convert Reddit discussions into bilingual Excel reports for competitor analysis, user insights, and market research.

carrielabs
User Research

UX Discovery Interviewer

AI-powered agent skill that conducts structured UX research interviews, transforming vague product ideas into actionable discovery insights.

JacobLinCool