docs
AI Assistants

AI Assistants Guide

Octigen uses specialized AI assistants throughout the application to help you create and manage presentations. Each assistant is optimized for specific tasks with appropriate tools and instructions.

Overview of Assistants

Template Assistant

Where: Template configuration page Purpose: Improve and configure templates Capabilities: Bulk shape renaming, slide categorization, usage classification, slide titling

Learn more about Template Assistant

Content Assistant

Where: Build Phase Purpose: Create presentation content Capabilities: Content creation, web research, table generation, brainstorming

Structure Analysis AI

Where: Structure Phase Purpose: Organize content into slide structure Capabilities: Content analysis, slide selection, structure generation

Slide Population AI

Where: Populate Phase Purpose: Fill slides with content Capabilities: Text generation, table creation, chart data generation

Refinement Assistant

Where: Refine Phase Purpose: Polish and perfect presentations Capabilities: Content modification, tone adjustment, slide management

Working with AI Assistants

General Principles

Be Specific

  • ✅ "Create a table showing Q1-Q4 revenue by product line with YoY growth percentages"
  • ❌ "Make a revenue table"

Provide Context

  • ✅ "For a technical audience, explain our API architecture focusing on scalability"
  • ❌ "Explain our API"

Iterate

  • First request might not be perfect
  • Refine with follow-up requests
  • Build incrementally

Trust But Verify

  • AI is powerful but not infallible
  • Always review generated content
  • Check facts and data
  • Verify appropriateness for audience

Conversation Patterns

Requesting Work:

You: "Create content for an executive summary highlighting our Q1 performance"
AI: [Creates content]
AI: "I've created an executive summary focused on Q1 performance. It highlights revenue growth, customer acquisition, and product launches."

Iterating:

You: "Make it more data-focused"
AI: [Updates content]
AI: "I've added specific metrics and percentages throughout."
You: "Good, now add a comparison to Q1 last year"
AI: [Updates again]

Getting Help:

You: "How should I structure a product launch presentation?"
AI: "For a product launch, I recommend this structure: 1) Problem statement, 2) Solution overview..."

Template Assistant

Specialization: Template structure and organization

Unique Capabilities

Bulk Operations:

  • Can modify many shapes/slides at once
  • Aggressive and thorough
  • Optimistic approach (doesn't ask permission)

Understanding:

  • Knows template structure
  • Tracks pending changes
  • Handles renamed shapes correctly

Best Workflows

"Categorize all slides based on their content"
"Rename all shapes with descriptive names"
"Mark all logos and page numbers as no touch"
"Give all slides descriptive titles"
"Find all shapes with default names and rename them"

Key Feature: Preview-and-Commit

Changes are staged for review before committing. You maintain full control.

Content Assistant

Specialization: Content creation and research

Unique Capabilities

Web Search:

  • Can search the internet for current information
  • Finds data, statistics, trends
  • Cites sources

Table Generation:

  • Creates formatted data tables
  • Organizes information effectively

Content Pieces Management:

  • Creates and updates content pieces
  • Main piece and additional pieces
  • Maintains structure

Best Workflows

"Research recent AI adoption trends and create a summary with statistics"
"Create content for a board presentation about our Q1 performance"
"Generate a comparison table of our pricing vs top 3 competitors"
"Expand the market opportunity section with more detail and data"
"Find recent news about [topic] and summarize key points"

Key Feature: Automatic Updates

Updates are automatically saved to content pieces. You see changes immediately in the editor.

Structure Analysis AI

Specialization: Content organization and slide selection

How It Works

  1. Analyzes your content from Build Phase
  2. Reviews available template slides
  3. Applies workflow rules (if applicable)
  4. Proposes slide-by-slide structure

Influencing Structure

Direct Feedback:

After first structure:
"Create more slides for the data sections"
"Use more visual slides"
"Reduce overall slide count"

Through Workflow:

  • Define deck structure
  • Set section requirements
  • Specify slide categories

Through Content:

  • Clear section breaks in content
  • Tables for quantitative slides
  • Visual descriptions for visual slides

Typical Results

  • Freeform: 8-15 slides for typical content
  • Workflow: Follows defined structure (5-20+ slides depending on definition)
  • Variety: Mixes slide types for visual interest
  • Flow: Logical progression from intro to conclusion

Slide Population AI

Specialization: Content generation for shapes

How It Works

For each slide:

  1. Determines what content should go on this slide
  2. Examines template slide shapes
  3. Generates appropriate content for each shape
  4. Applies tone and style preferences
  5. Formats appropriately

Influencing Population

Through Content:

  • Detailed, specific content = better population
  • Include exact data you want
  • Provide context and implications

Through Workflow:

  • Tone settings affect language
  • Style settings affect structure
  • Section instructions guide content

Through Template:

  • Shape names guide content type
  • Usage classification controls what's modified
  • Slide categories influence selection

Typical Output

  • Text boxes: Appropriate length, well-formatted
  • Tables: Data from your tables, formatted
  • Charts: Data converted to chart format
  • Titles: Descriptive and compelling

Refinement Assistant

Specialization: Interactive content adjustment

Unique Capabilities

Conversational:

  • Natural language requests
  • Explains what it's doing
  • Confirms changes

Precise Control:

  • Modify specific slides
  • Update specific elements
  • Fine-grained adjustments

Error Handling:

  • Graceful failure recovery
  • Explains issues
  • Retries automatically

Best Workflows

"Make slide 3 more detailed"
"Change the title on slide 5 to 'Q1 Financial Results'"
"Add a bullet point about our partnership to slide 7"
"Make the tone of slide 4 more formal"
"Remove slide 9"
"Add a new conclusion slide"

Key Feature: Immediate Feedback

Shows you what changed, confirms success, explains any issues.

Advanced Techniques

Chaining Requests

Build complex outcomes through multiple requests:

1. "Create an outline for a product launch"
2. "Expand section 3 with more technical details"
3. "Add customer testimonials to section 4"
4. "Create a pricing comparison table"
5. "Make the tone more persuasive overall"

Context Building

Provide rich context for better results:

"We're presenting to enterprise clients who are risk-averse and 
focused on ROI. Create content for a security overview section 
that addresses compliance, data protection, and enterprise-grade 
infrastructure. Include specific certifications we have."

Feedback Loops

Use AI output as starting point, then refine:

1. AI creates initial content
2. You review and identify gaps
3. "Add more detail about [specific aspect]"
4. AI enhances
5. "Now make it more concise"
6. AI condenses
7. Iterate until satisfied

Combining Assistants

Use different assistants for their strengths:

Template Assistant: Configure template perfectly Content Assistant: Create rich, detailed content Structure Analysis: Organize optimally Population AI: Generate complete slides Refinement Assistant: Polish to perfection

Common Pitfalls

Vague Requests

❌ "Make it better" ✅ "Make the language more technical and add specific performance metrics"

Too Many Changes at Once

❌ "Update slides 3, 5, 7, and 9 with new data and change the tone" ✅ "Update slide 3 with: [specific data]" → then next slide

Not Providing Enough Context

❌ "Write about our product" ✅ "Write about our cloud storage product, emphasizing security and compliance for healthcare industry, 250 words"

Expecting Perfection Immediately

  • AI provides strong starting points
  • Iteration is normal and expected
  • Multiple rounds produce best results

Not Verifying Output

  • Always check facts and data
  • Verify appropriateness
  • Review for errors

Best Practices

For All Assistants

  1. Be specific and detailed
  2. Provide context
  3. Iterate and refine
  4. Review and verify
  5. Learn from patterns

Maximizing Effectiveness

Understand Limitations:

  • Can't read your mind
  • Needs clear instructions
  • Makes mistakes sometimes

Play to Strengths:

  • Use for bulk operations
  • Leverage for research
  • Let it handle tedious tasks
  • Focus your time on strategy and review

Build Skills Over Time:

  • Learn what works
  • Develop request patterns
  • Understand each assistant's personality
  • Get faster with practice

Troubleshooting

Assistant Not Understanding:

  • Rephrase more simply
  • Break into smaller requests
  • Provide more context
  • Give examples

Unexpected Results:

  • Review your request
  • Check what you actually asked for
  • Clarify and try again
  • Provide corrective feedback

Errors or Failures:

  • Note error message
  • Try again (many errors are transient)
  • Simplify request
  • Contact support if persistent

Keyboard and Productivity Tips

Quick Access:

  • Chat sidebar usually on right
  • Click icon to open/close
  • Keyboard shortcuts (if available)

Efficient Workflow:

  • Keep chat open while working
  • Make requests as you notice issues
  • Don't batch too many changes
  • Verify each change before next

Suggested Prompts:

  • Most assistants offer suggested prompts
  • Click to insert
  • Great starting points
  • Learn effective patterns

Getting Help

Every assistant can help you understand:

  • What it can do
  • How to make requests
  • Why it made certain choices
  • What went wrong

Don't hesitate to ask:

"What can you help me with?"
"Why did you choose this slide for this content?"
"How should I structure this request?"
"What went wrong with my last request?"

Next Steps