Documentation Index
Fetch the complete documentation index at: https://docs.ephor.ai/llms.txt
Use this file to discover all available pages before exploring further.
- ELI5
- Deep Dive
Managing Chat Messages (Quick Take)
In a nutshell
Ephor shows newest messages first (top-down) and lets you pick the best AI response from multiple models to continue your conversation.
Why It Matters
Multiple AI models give different answers. Ephor lets you cherry-pick the best responses and build a conversation history from your favorite answers.Example
What is Ephor’s Message System?
What are Ephor Messages?
Ephor uses a top-down message flow with multi-model responses where you can select the best answer from different AI models to build your ideal conversation thread.
Why It Matters
Traditional chat is linear — one question, one answer. Ephor is multi-dimensional:- Compare perspectives — see how different AI models approach the same question
- Cherry-pick excellence — choose the best response from multiple options
- Build optimal conversations — create conversation history from your preferred answers
- Maintain context flow — starred responses carry forward seamlessly
You’re not just chatting with AI — you’re curating conversation excellence by selecting the best responses.
Message Structure & Flow
Top-Down Organization
Top-Down Organization
Newest messages appear first (unlike traditional chat):
- Top: Most recent messages and responses
- Bottom: Older conversation history
- Visual flow: Scroll up to see latest, down for context
- Why: Keeps current discussion prominent while preserving history
Message Couplets
Message Couplets
Each user message creates a response group:
- Your prompt: Single user query at the top
- AI responses: Multiple model responses grouped below
- Unified context: All responses reference the same conversation history
- Individual perspectives: Each model brings its unique approach
Continuous History
Continuous History
Despite multiple responses, context flows naturally:
- Starred responses (by double-clicking your preferred model) build the “official” conversation thread
- All models reference the same conversation history
- Switch between models mid-conversation without losing context
The Starred Response System
What are starred responses? Your selected “winning” answer that carries forward to the next message.How Response Selection Works
- Multi-Model Comparison
- Context Building
- Collaboration Impact
Each model brings unique strengths:
- Creative models: Imaginative, out-of-the-box thinking
- Analytical models: Data-driven, logical approaches
- Balanced models: Middle-ground perspectives
- Specialized models: Domain-specific expertise
Your conversation history is built from starred selections:
- Previous starred responses become context for new queries
- Models reference the same “official” conversation thread
- You can switch models mid-conversation seamlessly
- History remains coherent despite multi-model input
In shared and collaborative chats:
- Participant can star their preferred responses
- Everyone contributes to the shared conversation history
- Team decisions emerge from collective response curation
- Multiple perspectives strengthen the final conversation thread
Message Navigation Tips
Start from the top (newest) and scroll down for context. Think of it as reading a timeline in reverse-chronological order.
Don’t just pick the first good answer. Compare all responses — sometimes the best insights come from unexpected model perspectives.
Remember that your starred selections shape the entire conversation. Choose responses that best represent the direction you want the discussion to take.
Advanced Message Management
Strategic Response Selection
Strategic Response Selection
Think beyond individual answers:
- Opening moves: Choose responses that set good conversation tone
- Building complexity: Select answers that add depth progressively
- Maintaining focus: Star responses that stay on-topic and relevant
- Team alignment: In collaborative chats, consider group goals
Model Switching Strategy
Model Switching Strategy
Use different models for different purposes:
- Brainstorming: Start with creative models for idea generation
- Analysis: Switch to analytical models for evaluation
- Implementation: Move to practical models for execution planning
- Refinement: Use balanced models for final polishing
Conversation Quality
Conversation Quality
Maintain high-quality dialogue:
- Star responses that ask good follow-up questions
- Choose answers that build naturally on previous context
- Select responses that invite productive next steps
- Avoid starring responses that feel like dead ends