Skip to main content

​
Overview

A Project is an execution unit that separates knowledge by purpose. All resources (knowledge, episodes, chatbots) belong to a project.

​
Why Separate by Project

A single organization can have multiple purposes:
  • Product A support knowledge
  • Internal rules collection
  • Technical documentation
Mixing these into one knowledge base increases search noise and returns irrelevant information. Separating by project lets you build independent knowledge bases for each purpose.

​
Project Structure

Project
├── Settings
│   ├── Name & Description
│   └── LLM Model Configuration
├── Knowledge
│   ├── Knowledge (confirmed facts)
│   └── SourceContent (documents)
├── Episodes
│   ├── Episode (conversation & operation history)
│   └── Topic (episode summaries)
└── Chatbots
    └── Widget-based conversation system

​
Resource Ownership

All resources must belong to exactly one project.
ResourceDescription
KnowledgeConfirmed facts & entities
SourceContentDocument chunks
EpisodeConversation & operation history
TopicEpisode grouping & summary
ChatbotWidget-based conversation system
Resources cannot be moved between projects. To use in another project, you need to register them again.

​
Project Settings

​
Basic Settings

  • Name: Project identifier (e.g., “Product Support”)
  • Description: Notes and purpose description

​
LLM Model Settings

Configure which LLM models to use per project. Neuradex supports 6+ LLM providers, and you can use different models for different action types.

​
Supported Providers

ProviderKey Models
OpenAIGPT-4o, GPT-4o-mini
AnthropicClaude 4 Opus, Sonnet
GoogleGemini Pro, Flash
GroqOptimized for fast inference
FireworksOptimized inference
xAIGrok

​
Per-Action Model Configuration

ActionDescriptionRecommended
ChatLibrarian & Chatbot conversationsAccuracy-focused (GPT-4o, Claude)
EmbeddingKnowledge vectorizationHigh-quality embedding models
Knowledge ExtractionKnowledge generation from documentsStructured output-capable models
EnrichmentSummary & tag generationCost-efficient models
The optimal model varies by purpose. You can choose accuracy-focused models for technical documents or speed-focused models for FAQs based on project needs. Provider switching requires no code changes and takes effect immediately from the dashboard.

​
Ownership

Projects are owned by one of the following:
OwnerDescription
IndividualOnly that user can access
OrganizationOrganization members can access

​
Use Case Examples

​
Enterprise Use

Example Corp
├── Product Support (Project)
│   ├── FAQs, manuals as knowledge
│   └── Support chatbot deployed
├── Internal Wiki (Project)
│   ├── Company rules, procedures as knowledge
│   └── Librarian handles employee questions
└── Development Docs (Project)
    ├── API specs, design docs as knowledge
    └── Developer chatbot

​
Personal Use

Individual User
├── Reading Notes (Project)
│   └── Book summaries accumulated as knowledge
└── Learning Notes (Project)
    └── Learning content recorded as episodes

​
Next Steps

Librarian

Interact with the AI assistant

Data Types

Understand the four data types