Docs

AI Completions

Agentastic's AI completion feature brings the power of advanced language models directly to your fingertips, enabling intelligent text generation, rewriting, and assistance across all your applications.

What Are AI Completions?

AI completions go beyond simple text expansion by:

  • Understanding context and intent
  • Generating dynamic, relevant content
  • Adapting to your writing style
  • Providing intelligent suggestions
  • Helping with complex tasks

How AI Completions Work

  1. Trigger AI mode using special commands or the launcher
  2. Provide context through your prompt or selection
  3. AI processes your request using advanced language models
  4. Receive completion directly in your application
  5. Refine if needed with follow-up commands

Accessing AI Completions

Via Launcher

  1. Press ⌘+Space to open launcher
  2. Type your prompt or question
  3. AI responds in the overlay
  4. Insert response with ⏎ or copy with ⌘+C

Via Inline Commands

  1. Type /ai followed by your prompt
  2. Press space to trigger
  3. AI response replaces your text
  4. Continue typing normally

Via Selection

  1. Select existing text
  2. Open launcher with ⌘+Space
  3. Type instruction (e.g., "make this more formal")
  4. AI rewrites selected text

AI Models

Agentastic provides access to multiple AI models optimized for different tasks:

Fast Model (/model:fast)

  • Speed: <1 second response
  • Best for: Quick completions, simple tasks
  • Use when: Speed matters more than complexity

Smart Model (/model:smart)

  • Speed: 1-2 seconds
  • Best for: General writing, standard tasks
  • Use when: Balanced performance needed

Genius Model (/model:genius)

  • Speed: 2-5 seconds
  • Best for: Complex reasoning, creative tasks
  • Use when: Quality is paramount

Common AI Completion Tasks

Writing Assistance

Prompt: "Write a professional email declining a meeting"
Result: Formal, polite declination with alternatives

Code Generation

Prompt: "Python function to validate email addresses"
Result: Complete function with regex validation

Translation

Prompt: "Translate to Spanish: Thank you for your patience"
Result: "Gracias por su paciencia"

Summarization

Prompt: "Summarize this article in 3 bullet points"
Result: Concise, key-point summary

Grammar and Style

Prompt: "Fix grammar and make more concise"
Result: Corrected, streamlined text

Context-Aware Completions

Agentastic understands context from:

Current Application

  • Email clients: Professional tone
  • Code editors: Programming context
  • Chat apps: Casual style
  • Documents: Formal writing

Selected Text

  • Analyzes existing content
  • Maintains consistent style
  • Preserves formatting
  • Understands intent

Previous Interactions

  • Learns your preferences
  • Adapts to your style
  • Remembers context
  • Improves over time

Advanced AI Features

Multi-turn Conversations

Continue refining AI outputs:

  1. Get initial completion
  2. Request modifications
  3. Iterate until perfect
  4. Save as expansion for reuse

Vision Mode

Analyze visual content:

  1. Enable with /vision or the eye icon in launcher
  2. AI can see your screen
  3. Ask questions about visible content
  4. Get context-aware help

Chain Commands

Combine multiple AI operations:

/ai translate to French /model:smart /formal

Templates with AI

Create dynamic templates:

Trigger: weekly
Content: /ai generate weekly report for {{project}} including {{metrics}}

Prompt Engineering Tips

Be Specific

❌ "Write an email" ✅ "Write a friendly email thanking a colleague for their help with the project presentation"

Provide Context

❌ "Make this better" ✅ "Make this email more professional while maintaining a warm tone"

Specify Format

❌ "List benefits" ✅ "List 5 benefits as bullet points, each under 10 words"

Include Examples

❌ "Write like this" ✅ "Write in the same style as: 'We're excited to announce...'"

Privacy and Security

Local Processing

  • Prompts processed securely
  • No training on your data
  • Encrypted connections
  • Local caching only

Data Control

  • Clear cache anytime
  • Opt-out of analytics
  • No persistent storage
  • Your data stays yours

Sensitive Content

  • Automatic PII detection
  • Warning for sensitive data
  • Option to process locally
  • Compliance features

Performance Optimization

Response Speed

  • Use appropriate model
  • Keep prompts concise
  • Enable smart caching
  • Preload common requests

Quality Tips

  • Provide clear instructions
  • Use examples when helpful
  • Iterate on responses
  • Save good completions

Resource Usage

  • Automatic model selection
  • Efficient token usage
  • Background processing
  • Minimal memory impact

Integration with Other Features

With Text Expansion

  • Save AI outputs as expansions
  • Use AI to generate expansion content
  • Combine static and dynamic text

With Memory System

  • AI remembers your preferences
  • Contextual awareness improves
  • Personal writing style learned

With Special Tags

  • Combine /ai with other tags
  • Chain multiple operations
  • Create powerful workflows

Troubleshooting AI Completions

Slow Responses

  1. Check internet connection
  2. Try faster model
  3. Simplify prompt
  4. Clear cache

Unexpected Results

  1. Clarify prompt
  2. Add more context
  3. Try different model
  4. Use examples

Connection Issues

  1. Verify API access
  2. Check firewall settings
  3. Test with simple prompt
  4. Contact support

Best Practices

Daily Use

  1. Start with simple prompts
  2. Learn model strengths
  3. Build prompt library
  4. Share successful prompts

Professional Writing

  1. Always review AI output
  2. Maintain your voice
  3. Fact-check important content
  4. Use as starting point

Creative Tasks

  1. Experiment with models
  2. Try multiple variations
  3. Combine human and AI creativity
  4. Save inspiring outputs

Next Steps

Explore more AI capabilities:

AI completions transform Agentastic from a text expander into an intelligent writing assistant. The more you use it, the more valuable it becomes!