AI Chat for Research
Chat with multiple AI models to enhance your research, brainstorm ideas, and get instant assistance with your academic work.
Overview
The Chat feature gives you access to:
- Claude (Anthropic) - Expert at writing and analysis
- GPT-4 (OpenAI) - Strong at reasoning and creativity
- Gemini (Google) - Excellent for research and summaries
- 50+ other models - Including Mistral, Llama, and more
All conversations are project-specific and never used to train AI models.
Starting a Conversation
Step 1: Select Your Model
- Open the Chat tab in your project
- Click the Model Selector at the top
- Browse available models
- Click to select
- Model choice is saved for future messages
Quick Tips:
- Claude - Best for writing assistance and detailed explanations
- GPT-4 - Great for creative ideas and complex reasoning
- Gemini - Fast responses, good for quick summaries
- Try different models to find your preference
Step 2: Ask Your Question
- Type your question in the message box
- Press Enter or click Send
- Watch the response stream in real-time
- Follow up with related questions
Example Conversations
Example 1: Understanding a Paper
You: "What are the main findings in this quantum computing paper?"
Claude: [Provides clear summary with key points and implications]
You: "What methods did they use?"
Claude: [Explains the methodology in detail]
You: "How does this compare to previous work?"
Claude: [Provides context and comparisons]
Example 2: Brainstorming
You: "I'm researching machine learning in healthcare. What are interesting research directions?"
Claude: [Lists several promising areas with rationales]
You: "Tell me more about the diagnosis prediction angle"
Claude: [Dives deeper into that specific direction]
Example 3: Writing Help
You: "Help me write an abstract for my paper on... [full abstract text]"
Claude: [Provides feedback and suggestions]
You: "Make it more concise while keeping the key points"
Claude: [Rewrite with improvements]
Advanced Features
Thinking Mode
For complex questions, enable Thinking Mode:
- Toggle "Show thinking" switch before asking
- AI shows its reasoning process
- Get step-by-step analysis
- Perfect for methodology questions
When to use:
- Complex problem-solving
- Understanding intricate concepts
- Detailed analysis required
- Multi-step reasoning needed
Context Window
Each conversation has full context:
- AI remembers all previous messages
- Reference past points in the conversation
- Build on earlier ideas
- Progressively refine understanding
Tips:
- Start specific conversations for focused topics
- Create new conversations for new subjects
- Use conversation titles to stay organized
Model Capabilities
| Model | Strengths | Best For |
|---|---|---|
| Claude 3.5 | Writing, analysis, nuance | Essays, summaries, detailed feedback |
| GPT-4 | Reasoning, creativity, code | Complex problems, brainstorming, coding |
| Gemini | Speed, summaries, multimodal | Quick answers, paper summaries |
| Mistral | Efficiency, instruction following | Technical questions, code review |
| Llama | Open-source, good quality | Privacy-focused, cost-conscious |
Common Use Cases
1. Paper Summarization
Ask: "Summarize the methodology section of [paper title]"
Get: Clear breakdown of research methods
2. Methodology Discussion
Ask: "Is a qualitative or quantitative approach better for studying [topic]?"
Get: Detailed comparison with pros/cons
3. Literature Review
Ask: "What are the main themes in research about [topic] from 2020-2023?"
Get: Organized overview of key areas and papers
4. Writing Assistance
Ask: "Review this paragraph for clarity: [text]"
Get: Feedback and suggestions for improvement
5. Terminology
Ask: "Explain [technical term] in simple terms"
Get: Clear explanation with examples
6. Citation Help
Ask: "How should I cite [source] in APA format?"
Get: Properly formatted citation
7. Brainstorming
Ask: "What are novel research questions about [topic]?"
Get: List of interesting directions to explore
8. Code Review
Ask: "Review this code for errors: [code snippet]"
Get: Analysis and suggestions
Keyboard Shortcuts
- Cmd/Ctrl+L - Focus message input
- Shift+Enter - New line in message
- Cmd/Ctrl+K - Search conversations
- Up Arrow - Repeat last message (for quick edits)
Best Practices
Be Specific
❌ "Tell me about AI" ✅ "Explain how transformer models work in NLP"
Provide Context
❌ "Is this methodology good?" ✅ "I'm using surveys with 50 participants to study [topic]. Does this methodology work?"
Ask Follow-ups
Use the context to build on ideas:
- Initial question
- Follow-up for detail
- Another angle
- Refine the answer
Use Conversation Titles
- Name conversations for easy reference
- Helps you organize your research
- Easy to find later
Iterate and Refine
- First response not perfect? Ask clarifications
- Try different phrasing
- Combine answers from multiple models
Conversation Management
Organize Conversations
- Click "Rename" to give conversations meaningful titles
- Pin important conversations
- Archive old conversations
- Search conversation history
Export Conversations
- Open conversation
- Click "Export"
- Choose format: PDF or Markdown
- Save to your computer
Clear Conversation
- Click "Clear History"
- Confirm deletion
- Start fresh
Privacy & Data
Your conversations are:
- ✅ Stored securely on Citezy servers
- ✅ Never used to train models
- ✅ Encrypted in transit
- ✅ Only accessible to you
- ✅ Deleted when you delete them
Tips for Better Responses
- Give examples - Show what you mean
- Specify format - "Explain as a list" or "Write as a narrative"
- Add constraints - "Keep under 200 words" or "Assume beginner knowledge"
- Reference previous points - "Like you mentioned before..."
- Ask why - Deepen your understanding
Model Switching
You can switch models mid-conversation:
- Select different model from dropdown
- Type new question
- New model responds with conversation context
- Try multiple models on same topic
Why do this?
- Different models have different strengths
- Compare perspectives on the same question
- Find the best model for your task
Common Questions
Q: Is my data used to train models? A: No. All models have training disabled for your data.
Q: Can I use these conversations in my paper? A: Yes, but cite appropriately. Example: "Personal communication with Claude AI, 2024"
Q: Which model is best? A: Try them all! Each excels at different tasks.
Q: How long can conversations be? A: Very long. AI remembers full context, but shorter focused conversations work better.
Q: Can I share conversations? A: Yes, export them and share PDFs/Markdown with collaborators.
Limitations to Know
- AI can make mistakes - verify important facts
- Outdated knowledge (training data has cutoff date)
- Can't access live web or your uploaded files
- Shouldn't be sole source for critical decisions
Next Steps
- Manage your references alongside chat
- Search papers and discuss them in chat
- Learn about AI detection to identify AI content