BMAD-METHOD/src/modules/creative-prompting-studio/workflows/research-workflow/workflow.md

Research Workflow - Perplexity Prompting

Agent: Perplexity Specialist Purpose: Create comprehensive, well-structured research prompts for Perplexity AI

Overview

This workflow guides you through creating effective research prompts using the Perfect Prompt Framework, ensuring you get accurate, well-cited, and comprehensive information from Perplexity AI.

When to Use

• Academic research and literature reviews
• Market research and competitive intelligence
• Current events and trend analysis
• Business intelligence and industry analysis
• Product research and user needs
• Technical research and documentation
• Any information gathering that requires real-time, cited sources

Workflow Steps

Step 1: Define Research Need

Goal: Clarify what you need to know and why

Actions:

1. State your research question or information need
2. Define the context and intended use
3. Identify what would constitute a successful answer
4. Determine appropriate depth level

Questions to Answer:

• What specific question needs answering?
• What will you do with this information?
• What depth do you need (quick facts, standard, comprehensive, academic)?
• Are there time constraints (current events vs. timeless knowledge)?

Command: analyze-research-need

Step 2: Choose Prompt Category

Goal: Select the right approach for your research

Categories:

• Informational - Gathering facts, data, and current information
• Instructional - How-to guides and step-by-step processes
• Interactive - Expert consultation and comparative analysis

Considerations:

• Informational: "What is...?", "Show me current data on..."
• Instructional: "How do I...?", "What are the steps to...?"
• Interactive: "As an expert in X, analyze...?", "Compare A and B..."

Step 3: Build Core Prompt

Goal: Create the foundation using the 5-component framework

Components to Define:

1. Clear Instruction - What you want Perplexity to do

   • Be specific and actionable
   • Use clear verbs (analyze, summarize, compare, find, explain)

2. Relevant Context - Background information

   • Industry, field, or domain context
   • Why you're asking (use case)
   • Any constraints or assumptions

3. Specific Input - The subject matter

   • Precise topic or data to analyze
   • Relevant time frames
   • Geographic or domain boundaries

4. Key Keywords - Focus terms

   • 3-5 critical search terms
   • Specific terminology for your domain
   • Alternative terms to consider

5. Output Format - How results should be structured

   • List, table, narrative, comparison matrix
   • Sections or categories needed
   • Level of detail for each section

Command: create-research-prompt or template-specific commands

Step 4: Add Specialization

Goal: Tailor the prompt for your specific use case

Professional Templates Available:

• Educational - Curriculum, learning resources, topic exploration
• Product Management - Market research, user needs, competitive analysis
• Financial Analysis - Market trends, company research, economic indicators
• Marketing Intelligence - Audience insights, competitive tracking, trends

Enhancement Options:

• Request specific citation formats (APA, MLA, Chicago)
• Ask for source diversity (academic, industry, news)
• Require recency (within last 6 months, current year)
• Specify expertise level (general audience, expert level)

Commands: template-educational, template-product-management, etc.

Step 5: Optimize for Perplexity

Goal: Leverage Perplexity's unique strengths

Optimizations:

• Real-time Data: Request current information, latest trends
• Source Verification: Ask for credible, diverse sources
• Live Web Search: Emphasize need for most recent data
• Citation Requirements: Specify how sources should be referenced

Add to Prompt:

• "Include citations for all claims"
• "Focus on sources from the last [timeframe]"
• "Provide a mix of academic and industry sources"
• "Highlight any conflicting information from different sources"

Step 6: Review Against Checklist

Goal: Ensure prompt quality and completeness

Quality Checklist:

• ✅ Instructions are clear and unambiguous
• ✅ Context is sufficient for understanding
• ✅ Inputs are specific and well-defined
• ✅ Keywords are relevant and focused
• ✅ Output format is specified
• ✅ Scope is appropriate and manageable
• ✅ Real-time/current data is requested
• ✅ Citation requirements are included
• ✅ Language is straightforward, not vague

Common Issues to Fix:

• Vague terms: "good", "best", "nice" → Replace with specific criteria
• Too broad: Narrow scope or break into multiple prompts
• Missing format: Add structure for how results should appear
• No citations: Always request sources and verification

Command: refine-research-prompt

Step 7: Test and Iterate

Goal: Refine based on actual results

Process:

1. Use the prompt in Perplexity
2. Evaluate the results:
   • Did it answer your question?
   • Is the depth appropriate?
   • Are sources credible and well-cited?
   • Is the format usable?
3. Identify gaps or issues
4. Refine specific components
5. Test again if needed

Iteration Focus Areas:

• Adjust depth if too superficial or too detailed
• Add constraints if results are too broad
• Refine keywords if missing key information
• Modify output format if structure isn't helpful

Output Deliverables

Research Prompt Document

Saved to: {{research_prompts_folder}}/[topic]-research-prompt.md

Contents:

# Research Prompt: [Topic]

## Purpose
[What you're researching and why]

## Perplexity Prompt

[Complete, ready-to-use prompt]

## Metadata
- Category: [Informational/Instructional/Interactive]
- Depth: [Quick/Standard/Comprehensive/Academic]
- Template: [Professional template used, if any]
- Created: [Date]
- Status: [Draft/Tested/Refined]

## Expected Outputs
- [What you expect to receive]
- [Key questions that should be answered]

## Notes
[Any refinements needed, lessons learned, etc.]

Success Criteria

A successful research workflow produces:

• ✅ Clear, unambiguous prompt ready to use
• ✅ All 5 framework components present and optimized
• ✅ Appropriate depth for your needs
• ✅ Citation and source requirements specified
• ✅ Output format defined and usable
• ✅ Leverages Perplexity's real-time capabilities
• ✅ Documented for future reference and iteration

Tips for Success

1. Start Specific: Better to be too narrow and expand than too broad
2. Request Citations: Always ask for sources and verification
3. Define Format Early: Knowing output structure helps focus the search
4. Use Templates: Leverage professional templates as starting points
5. Iterate: First prompts rarely perfect; refine based on results
6. Document: Save successful prompts for similar future needs
7. Think Real-Time: Perplexity's strength is current info; use it
8. Be Clear About Depth: Specify how comprehensive you need results

Common Use Cases

Quick Fact Checking

• Minimal context needed
• Specific, targeted questions
• Request verification from multiple sources

Market Research

• Use Product Management or Marketing templates
• Request competitive landscape
• Ask for recent trends and data

Academic Research

• Use Academic template
• Specify citation format
• Request scholarly and peer-reviewed sources
• Ask for recent studies and literature

Competitive Intelligence

• Request real-time competitor information
• Ask for market positioning analysis
• Require multiple source verification
• Include industry context

Learning & Education

• Use Educational template
• Request multiple perspectives
• Ask for examples and applications
• Specify expertise level

---

Master the art of research prompting to unlock Perplexity's full potential for accurate, comprehensive, and well-cited information.