Advanced Workflow Using AI Instruction System

Overview

By leveraging the diverse instruction files in the AI instruction system, you can streamline the entire presentation creation process with AI assistance.

Phase-Based Workflow

1. Research & Analysis Phase

Data Collection and Analysis

# Load data analysis instructions
"Please load `.ai-instructions/instructions/ja/analysis/basic_data_analysis.md`"

# Example analysis request
"Analyze the market research data and create graphs showing the trend of AI tool adoption rates"

Technical Research

# Research as Python expert
"Please load `.ai-instructions/instructions/ja/agent/python_expert.md`"

# Research the latest ML frameworks and create a comparison table

2. Idea Generation Phase

Brainstorming

# Utilize creative instructions
"Please load `.ai-instructions/instructions/ja/creative/basic_creative_work.md`"

# Propose 10 catchphrases for the presentation

Multi-perspective Consideration

# Consider from different perspectives
"From the perspectives of engineers, executives, and end-users, 
propose key points to emphasize in this presentation"

3. Structure Design Phase

Storyline Creation

# Structure as technical writer
"Please load `.ai-instructions/instructions/ja/agent/technical_writer.md`"

# Propose a structure that explains technical content clearly to non-technical audiences

4. Content Creation Phase

Slide Creation

# Streamline with basic code generation instructions
"Please load `.ai-instructions/instructions/ja/coding/basic_code_generation.md`"

# Create slides including Mermaid diagrams and code examples

Quality Check

# Improve quality with code review instructions
"Please load `.ai-instructions/instructions/ja/agent/code_reviewer.md`"

# Review the code examples in the slides and suggest improvements

Practical Example: Creating a Technical Presentation

Step 1: Market Research

# Analyze market data with data analysis instructions
"Analyze AI tool market growth rate data and create visually clear graphs"

Step 2: Technical Comparison

# Technical evaluation as Python expert
"Create a performance comparison table of major AI frameworks and summarize their features"

Step 3: Message Design

# Generate ideas with creative instructions
"Propose ideas for an impressive opening slide with the theme 'The Future of Development Changed by AI'"

Step 4: Content Writing

# Write as technical writer
"Create slides that explain technical content in a way executives can understand"

Step 5: Final Check

# Overall review
"Review the entire presentation and suggest improvements"

Checkpoint Utilization

Progress in each phase is automatically recorded in checkpoint.log:

[2025-01-10 10:00:00] [TASK-a1b2c3] [START] Starting technical presentation creation (estimated 5 steps)
[2025-01-10 10:30:00] [TASK-a1b2c3] [COMPLETE] Outcome: Market research data analysis completed, 3 types of graphs created
[2025-01-10 11:00:00] [TASK-a1b2c3] [START] Technical comparison research
[2025-01-10 11:30:00] [TASK-a1b2c3] [COMPLETE] Outcome: 5 framework comparison table created
[2025-01-10 14:00:00] [TASK-a1b2c3] [START] Creating presentation first draft
[2025-01-10 15:00:00] [TASK-a1b2c3] [COMPLETE] Outcome: 20 slides, 3 demos completed

Integrated Script Utilization

Using create-presentation.sh, you can build a consistent environment from research to presentation:

Creating a Full Project

# Create complete structure with research, ideas, and slides
./scripts/create-presentation.sh --full research-presentation

# Manage as GitHub repository (with CI/CD)
./scripts/create-presentation.sh --github --full tech-conference-2025

Created Structure (Full Project)

research-presentation/
├── research/          # Used in research phase
│   ├── notes.md      # Record research content with AI assistance
│   ├── data/         # Store raw data
│   └── analysis/     # Output analysis results
├── ideation/         # Idea generation phase
│   ├── brainstorm.md # Develop ideas with AI assistance
│   └── drafts/       # Structure drafts
├── assets/           # Visual materials
│   ├── images/       # Image materials
│   └── charts/       # Graphs and diagrams
└── slides.md         # Final presentation

Tips

Combining Instructions

Utilize different instructions in the flow: research → analysis → creation → writing
Review the same content from different perspectives

Efficient Switching

# Work while switching instructions
"First analyze market data with the analysis instructions, 
then generate visual ideas with the creative instructions"

Quality Improvement Tips

Select appropriate expert-type instructions for each phase
Regularly check progress with checkpoints
Finally review from multiple perspectives

By utilizing the AI instruction system and integrated scripts this way, presentation creation can be managed as a comprehensive project.