Quick Start Guide

What is Workunit?

Workunit is a multi-LLM context sharing platform. Create a workunit once, then use Claude, GPT, Gemini, or any AI model to work on it—all sharing the same context and real-time task status.

The Problem: You use Claude for planning, GPT for execution, and Gemini for analysis—but each model starts from zero. You waste hours re-explaining context.

The Solution: Create a workunit once. Every AI model can access it, see what other models have done, and update task status in real-time.

5-Minute Start

Follow these steps to create your first workunit and connect AI assistants. You'll be productive within 10 minutes.

Step 1: Your Organization

Every workunit belongs to an organization. If you just signed up, you'll be prompted to create or join one.

Solo founders: Create an organization for yourself
Teams: Create one organization for your team
Joining others: Use an invite link from your team

Step 2: Create Your First Workunit

A workunit represents a unit of work with a clear problem and success criteria. Click 'New Workunit' and fill in:

Name

Clear, concise description of what you're building

Example: "Build user authentication system"

Problem Statement

What problem are you solving? Why does this matter?

Example: "Users need secure login and registration. Current system has no authentication, exposing data and functionality to unauthorized access."

Success Criteria

How will you know when this is complete?

Example: "Users can register with email/password, login securely, logout, and access protected resources. All passwords hashed with bcrypt. JWT tokens expire after 24 hours."

Priority and tags are optional but help with organization. Start with 'normal' priority and 'feature' tag.

Step 3: Add Tasks

Break your workunit into 2-5 actionable tasks. Keep it simple—you can add more tasks later.

Example Tasks for Authentication:

Design database schema for users table

Implement registration endpoint with bcrypt password hashing

Implement login endpoint with JWT token generation

Add authentication middleware to protected routes

Write integration tests for auth flows

Step 4: Connect AI Assistants

Connect your AI tools to Workunit using Model Context Protocol (MCP). This takes 30 seconds.

For Claude Code:

Run this command in your terminal:

claude mcp add --transport http workunit https://workunit.app/mcp

Restart Claude Code after installation

For Gemini CLI:

Run this command in your terminal:

gemini mcp add --transport http workunit https://workunit.app/mcp

Then authenticate: /mcp auth workunit

MCP Integration Guide

See setup instructions for other tools and troubleshooting help

Step 5: Use Workunit from AI

Now any AI model can access your workunit. Here's what to ask:

Read your workunit:

"Show me my active workunits"

Get full context by name:

"Get workunit 'JWT Authentication' with AI context and tasks"

Or use the direct URL (easier!):

"Get this workunit: https://workunit.app/workunits/a1b2c3d4-e5f6-7890-abcd-ef1234567890"

Pro tip: Just copy-paste the URL from your browser! No need to remember names or IDs.

Update task status:

"Mark task 'Design database schema' as in progress"

Add AI context:

"After implementing the schema, update the workunit with what I learned"

Practical Example: Multi-Model Workflow

See how to use multiple AI models on the same workunit, each leveraging their strengths:

Claude Sonnet 4.5 - Planning Phase

You to Claude:

"Create a plan for building user authentication. Include security best practices."

Claude:

Creates detailed workunit with security considerations, breaks down into 5 tasks, adds AI context about authentication patterns and JWT security.

GPT-5 - Implementation Phase

You to GPT:

"Get workunit 'Build user authentication' and implement the database schema task"

GPT:

Reads Claude's plan, implements schema with proper indexes and constraints, marks task complete, adds implementation notes to AI context.

Gemini - Code Review Phase

You to Gemini:

"Review workunit 'Build user authentication' - analyze the implementation for security issues"

Gemini:

Reads Claude's plan and GPT's implementation, performs security analysis, suggests bcrypt cost factor adjustment, updates AI context with findings.

Result: Each model used its strengths while building on previous work. No context lost, no duplicate effort, no re-explaining.

Next Steps

You're ready to use Workunit! Here's what to explore next:

Core Concepts

Understand workunits, tasks, assets, and how context flows between AI models

MCP Integration

Complete guide to connecting Claude, Gemini, and other AI tools via Model Context Protocol

Team Collaboration

Invite team members, share workunits, and collaborate with both humans and AI

Asset Management

Track products, systems, people, and knowledge that your AI models need to understand

Need Help?

Join our community, read the docs, or reach out to support. We're here to help you succeed with multi-model AI workflows.

Documentation Community Discussions