Skip to main contentWhat Are Custom Models?
Simple Explanation
Custom models are personalized AI assistants that combine a base model with your own instructions and knowledge base. They use RAG (Retrieval Augmented Generation) to access your specific data and provide more personalized, accurate responses based on your information.
Key Components
Base Model:
- Choose any AI model available in the app as your base
- The base model determines the AI’s core capabilities and style
- You can switch base models to find the best fit for your needs
Knowledge Base:
- Upload documents, files, or data you want the AI to reference
- The AI uses RAG to retrieve relevant information from your knowledge base
- Perfect for company documents, research papers, personal notes, or specialized content
Instructions:
- Provide custom instructions to guide how the AI responds
- Define the AI’s role, tone, and behavior
- Specify how it should use the knowledge base
Privacy Settings:
- Make your custom model public to share with others
- Keep it private for personal or internal use only
When to Create One
- Company Knowledge Base: Create an AI assistant that knows your company documents and policies
- Personal Assistant: Build an AI that understands your preferences and information
- Specialized Domain: Create assistants for specific topics (medical, legal, technical)
- Research Assistant: Upload research papers and get answers based on your sources
- Educational Content: Build assistants from textbooks or course materials
Creating a Custom Model
Getting Started
Currently Available:
- Web platform only
- Mobile support coming soon
On Web:
- Navigate to Custom Models section
- Click “Create New Custom Model”
- Choose a base model from available options
- Upload files or documents for your knowledge base
- Add custom instructions to define behavior
- Set privacy (public or private)
- Save and start using your custom model
Choosing a Base Model
Select any model available in the app as your base.
The base model you choose affects how your custom model processes and responds to queries.
Building Your Knowledge Base
Your knowledge base is the data your custom model will reference:
Supported Content:
- Documents (PDF)
- Multiple files can be combined
Best Practices:
- Organize related documents together
- Use clear, well-structured content
- Include relevant context and examples
- Keep knowledge base updated as needed
The AI uses RAG to search through your knowledge base and retrieve the most relevant information when answering questions.
Writing Instructions
Custom instructions guide how your model behaves:
What to Include:
- The AI’s role and purpose
- How it should respond (tone, style, format)
- Any specific guidelines or constraints
Example Instructions:
- “You are a technical support assistant. Use the knowledge base to answer questions about our product. Be concise and helpful.”
- “You are a research assistant. Reference the uploaded papers when answering. Always cite sources.”
Privacy Settings
Private:
- Only you can access and use the model
- Perfect for personal assistants or internal company use
Public:
- Share your custom model with others
- Others can use it but cannot modify it
- Great for sharing specialized assistants with your team or community
Using Custom Models
Once created, you can use your custom model just like any other AI model:
- Select your custom model from the model selector
- Start chatting - the AI will use your knowledge base and instructions
- Ask questions that relate to your knowledge base
- The AI retrieves relevant information and provides personalized answers
Tips for Better Results
- Clear Instructions: Be specific about the AI’s role and how it should behave
- Quality Knowledge Base: Use well-organized, relevant documents
- Appropriate Base Model: Choose a base model that matches your use case
- Regular Updates: Keep your knowledge base current with new information
- Test and Refine: Try different instructions and base models to find what works best
