Project Setup
Configure your RAG project name, domain, and use case in Step 1 of the wizard.
The first step of the RAG Wizard establishes the foundation for your RAG system. The choices you make here influence chunking recommendations, embedding model selection, and retrieval strategy.
Accessing the Wizard
Navigate to Dashboard → RAG → Create New to start the wizard.
Form Fields
Project Name
A unique identifier for your RAG project. This name appears in the dashboard, API responses, and logs.
| Parameter | Type | Default | Description |
|---|---|---|---|
Length | 1-100 characters | — | Must be unique across your account. |
Allowed chars | Letters, numbers, hyphens, underscores | — | No spaces or special characters. |
Good examples:
customer-support-kbproduct-docs-2024legal-contracts-archive
Project names cannot be changed after creation. Choose a name that will remain relevant as your project grows.
Project Description
A brief summary of what this RAG system does. This helps your team understand the purpose and assists the wizard in recommending optimal settings.
Example:
Knowledge base for customer support team. Contains product manuals,
FAQ documents, and troubleshooting guides for our SaaS platform.
Expected to handle 500+ queries per day from support agents.
Domain Selection
Choose the domain that best matches your use case. This helps optimize chunking and retrieval strategies.
| Domain | Best For | Example Use Cases |
|---|---|---|
| Customer Support | FAQ, help desk | Product manuals, troubleshooting guides |
| Technical Documentation | API docs, code | Developer guides, architecture docs |
| Legal & Compliance | Contracts, policies | Terms of service, regulations |
| Research & Academia | Papers, citations | Literature reviews, research synthesis |
| Business Intelligence | Reports, analysis | Sales data, market research |
| Healthcare | Medical knowledge | Clinical guidelines, patient info |
| Education | Course materials | Lectures, study guides |
| General | Mixed content | Personal knowledge base, misc docs |
Use Case Description
Provide specifics about how your RAG system will be used. This helps the wizard recommend:
- Optimal chunk size for your query patterns
- Appropriate embedding model
- Retrieval method (dense, sparse, hybrid, graph)
Example descriptions:
| Use Case | Description |
|---|---|
| FAQ Bot | "Answer customer questions about product features and pricing from our help docs" |
| Code Assistant | "Find relevant code examples and API documentation for developers" |
| Contract Review | "Search legal contracts for specific clauses and compliance requirements" |
| Research Assistant | "Synthesize findings from academic papers on machine learning" |
LLM Integration
Choose whether to use LLM-generated answers alongside retrieved chunks.
| Option | Description | When to Use |
|---|---|---|
| Chunks only | Return raw matching chunks | When you want full control over response generation |
| LLM answer | Generate answers from retrieved context | When you want natural language responses |
Start with "Chunks only" to verify retrieval quality before enabling LLM answers. You can always enable LLM later from the Pipeline Configuration step.
Validation
The wizard validates the following before allowing you to proceed:
- Project name is unique
- All required fields are filled
- Domain and use case are selected
Next Step
After completing Project Setup, you'll move to Data Sources to upload your documents.
- Write a detailed description — it helps with recommendations
- Choose the domain that matches your content type
- Be specific about query patterns in use case
- Use generic names like "test" or "project1"
- Leave the description blank
- Select "General" if a more specific domain fits