Orchestrator
The Orchestrator is your intelligent interface to the entire Mantis platform, making complex data analysis and visualization tasks accessible through simple conversation. It acts as a powerful assistant that can read your documents, perform web searches, generate visualizations, and control the interface—all through natural language commands.
What the Orchestrator Does
Document Understanding & Analysis
- Reads and understands your uploaded documents
- References specific documents in responses with citations like
[2:doc_id] - Ranks documents by relevance to your query
- Provides source URLs when available
Real-Time Information
- Performs web searches when you need current information
- Fetches and processes web content
- Integrates live data with your existing documents
Smart Command Generation
- Creates visualization commands (plots, charts, graphs)
- Controls interface panels (analysis, chat, maps, etc.)
- Manages tools and view modes
- Handles data selection and filtering
What You Can Use It For
Data Visualization
Transform natural language into powerful visualizations:
- "Create a scatter plot of sales vs marketing spend" → Automatically generates the right plotting commands
- "Show me a bar chart of monthly revenue" → Creates the appropriate visualization
- "Generate a heatmap of customer satisfaction scores" → Builds complex visualizations
Research & Analysis
Get current information and insights:
- "Find the latest research on machine learning trends" → Searches the web and summarizes findings
- "What are the current market trends in our industry?" → Fetches real-time data and analysis
- "Compare our performance with industry benchmarks" → Gathers and analyzes comparative data
Document Analysis
Extract insights from your uploaded materials:
- "Summarize the key points from my uploaded reports" → Reads your documents and provides insights
- "Find all mentions of 'customer feedback' in my documents" → Searches through your content
- "What are the main themes in my research papers?" → Analyzes patterns and themes
Interface Control
Navigate the platform effortlessly:
- "Open the analysis panel and show me the data inspector" → Navigates the interface for you
- "Switch to the map view and zoom in on the northeast region" → Controls visualization modes
- "Show me the quantitative analysis tools" → Opens relevant panels and tools
Complex Workflows
Execute multi-step processes:
- "Compare Q1 vs Q2 sales, create a visualization, and highlight the differences" → Breaks this into multiple coordinated steps
- "Analyze customer sentiment, create a report, and export the findings" → Orchestrates complex workflows
- "Find outliers in our data, create visualizations, and prepare a summary" → Handles end-to-end analysis
How-To
Step-by-Step Guide
Open the Orchestrator Interface
- Navigate to the Orchestrator tab in your workspace
- You'll see a chat interface with a text input field
Select Your Agent Mode
- Choose from the dropdown menu:
- Narrator: For storytelling and narrative responses
- Agent: For general AI assistance
- Simple Coder: For code generation and technical tasks
- Choose from the dropdown menu:
Enter Your Request
- Type your natural language command in the input field
- Be specific about what you want to achieve
- Include relevant context about your data or goals
Review and Refine
- The Orchestrator will process your request and provide results
- If needed, ask for refinements or clarifications
- Use the "Clear Context Points" button to reset if necessary
Iterate and Improve
- Provide feedback on the results
- Ask follow-up questions to refine the output
- Use the "Add Context" button to provide additional information
Interface Controls
- Clear Context Points: Reset the conversation context
- Enable Visual Brainstorming: Activate visual thinking tools
- Add Context: Provide additional files or information
- Chat: Send your message and get a response
- Clear: Clear the current input or conversation
- Agent: Switch between different AI agent modes
How It Works
1. Input Processing
You send a natural language request through the chat interface. The Orchestrator understands context, intent, and specific requirements.
2. Context Gathering
The system considers:
- Your uploaded documents and their content
- Current workspace state and data
- Previous conversation history
- Available tools and capabilities
3. Command Generation
Your request is converted into specific platform commands:
- Visualization commands for charts and graphs
- Interface navigation commands
- Data analysis and filtering operations
- Web search and information gathering tasks
4. Execution
The Orchestrator runs the generated commands and provides results:
- Creates visualizations in the appropriate panels
- Navigates to relevant interface sections
- Performs data analysis and filtering
- Fetches and processes external information
5. Iteration
If needed, the system can refine and improve results:
- Adjusts visualizations based on feedback
- Refines search queries for better results
- Optimizes analysis parameters
- Provides alternative approaches
Getting Started
Basic Usage
- Open the Orchestrator through the chat interface
- Describe what you want in natural language
- Review the results and provide feedback if needed
- Iterate and refine until you get the desired outcome
Best Practices
- Be specific about what you want to achieve
- Mention data sources when relevant (e.g., "using my sales data")
- Specify visualization types if you have preferences (e.g., "create a line chart")
- Ask for refinements if the initial result isn't quite right
Example Commands
"Create a scatter plot showing the relationship between marketing spend and sales revenue"
"Find the latest research papers on artificial intelligence in healthcare"
"Open the analysis panel and show me the top 10 performing products"
"Compare our Q1 and Q2 performance and create a side-by-side comparison"
"Search the web for current trends in our industry and summarize the findings"Integration with Other Features
The Orchestrator works seamlessly with all other Mantis features:
- Exploration tools for data discovery
- Inspection tools for detailed analysis
- Quantitative tools for statistical analysis
- Machine learning features for advanced modeling
- All visualization and mapping capabilities