AI Narrative Generation
How AccredLeap uses AI to generate accreditation document narratives from your institutional data.
How AI Generation Works
AccredLeap uses Claude, Anthropic's AI model, to transform your interview responses and institutional data into polished accreditation narratives. The AI is specifically prompted with accreditation standards knowledge to produce content that meets the expectations of peer reviewers.
Generating a Section
- Open a document and navigate to the section you want to generate
- Ensure you have completed the relevant interview sections (the system will indicate readiness)
- Click Generate on the section
- The AI will produce a draft narrative based on your interview data and the section's accreditation requirements
Regeneration & Refinement
After the initial generation, you can refine the output using the regeneration panel:
Adjustment Options
- Make it shorter — Condense the narrative while preserving key points
- Make it longer — Expand with additional detail and context
- More formal tone — Shift toward academic/institutional language
- More evidence-focused — Emphasize data points and supporting evidence
- Custom instruction — Provide specific guidance for the AI (e.g., "Emphasize our new curriculum changes from 2024")
How Regeneration Works
Each regeneration request sends your original interview data plus the current narrative and your adjustment instructions to the AI. The system preserves context so refinements build on previous iterations rather than starting from scratch.
Collaborative Editing
After generation, the narrative appears in a rich text editor where your team can:
- Make direct edits to the AI-generated text
- Leave inline comments for reviewers
- Track changes across editing sessions
- Edit collaboratively in real-time with other team members
Data Privacy
- Your institutional data is processed securely through Supabase Edge Functions
- Interview responses and generated content are stored in your institution's isolated database
- AI prompts are not used for model training
- All data transmission is encrypted in transit
Next Steps
- Approval Workflows — Route completed sections for review
- Getting Started — Return to the overview