AI Compliance Notes
The AI Compliance Notes feature helps vCIOs and technicians quickly generate professional, client-facing public notes from their technical private notes using OpenAI’s GPT models.Overview
When working on compliance checks, technicians often need to maintain two sets of notes:- Private notes: Technical details, internal observations, and detailed remediation steps
- Public notes: Professional, client-facing summaries written in business language
- Use clear business language without technical jargon
- Focus on business impact and value
- Provide actionable remediation steps
- Maintain a professional tone without unnecessary pleasantries
Prerequisites
Before using the AI compliance notes feature, ensure:- You have private notes written for the compliance check
- You have permission to edit compliance checks
Using AI Draft
Generating Public Notes
- Open a compliance check dialog
- Navigate to the Notes section
- Enter your technical observations in the Private Notes field
- Click the AI Draft button next to Public Notes
- The AI will generate professional public notes that stream in real-time
- Review and edit the generated content as needed
- Save the compliance check
The AI Draft button is only enabled when private notes are present, as it uses them as the source for generating public notes.
Generated Content Format
The AI generates notes with:- Clear headers using Markdown formatting
- Bullet points for easy scanning
- Structured sections for issue summary, impact, and remediation
- Concise content limited to approximately 120 words
- Professional tone without filler phrases or pleasantries
Example Output
Given private notes about a firewall configuration issue, the AI might generate:Customizing the AI Behavior
The AI assistant is configured to:- Generate content suitable for client communication
- Focus on business value and impact
- Avoid technical jargon
- Maintain a professional, direct tone
- Limit output to approximately 1000 words
Best Practices
Writing Effective Private Notes
To get the best results from AI generation:- Be specific about technical issues found
- Include context about why issues matter
- Document remediation steps you plan to take
- Note any risks or impacts to the business
Reviewing Generated Content
Always review AI-generated content before saving:- Verify accuracy of technical details translated to business language
- Ensure completeness of recommendations
- Check tone is appropriate for the specific client
- Add client-specific context where needed
- Remove any hallucinated information that doesn’t match your findings
Limitations
- Maximum output length of approximately 1000 words
- Requires private notes to generate content
- Generated content should always be reviewed before client delivery
- AI may not capture client-specific terminology or preferences
- Real-time streaming requires stable internet connection
Troubleshooting
AI Draft Button Disabled
If the AI Draft button is disabled:- Ensure you have entered private notes
- Check that you have edit permissions for the compliance check
- Verify the check is not in historical/read-only mode
Generation Fails
If content generation fails:- Ensure your internet connection is stable
- Try reducing the length of your private notes
Content Not Appearing
If generated content doesn’t appear immediately:- The content streams in real-time - wait a few seconds
- Check for any error messages in the toast notification
- Refresh the dialog if the issue persists
Technical Details
Architecture
The AI compliance notes feature uses:- Server Actions for secure API calls
- Streaming responses for real-time content display
- Markdown to HTML conversion for rich text formatting
- Optimistic updates for immediate UI feedback
Security
- OpenAI API calls are made server-side only
- API keys are never exposed to the client
- Content is streamed through secure WebSocket connections
- All generated content is saved to your secure database
Performance
- Streaming typically begins within 1-2 seconds
- Complete generation takes 3-5 seconds for 120 words
- Content updates are debounced to prevent excessive API calls
- Generated content is cached until the dialog is closed