Skip to main content
Version: 1.0.0-beta

Troubleshooting — AI

AI features are not available

  • Check that an AI model is configured in Administration → AI Models
  • Verify the model connection test passes
  • Ensure your role has ai:suggest permission (Auditor, Manager, CAE)

AI responses are slow

  • Large documents take longer to process
  • Check the AI model provider status
  • For Ollama: ensure the local server has sufficient resources
  • Try a smaller/faster model for routine tasks

AI generates inaccurate content

  1. Provide more context (scope, objectives, detailed findings)
  2. Be more specific in your prompts
  3. Check the confidence score — low scores indicate uncertainty
  4. Try regenerating with additional context
  5. Consider switching to a more capable model

AI can't find historical data

  • RAG indexing may be in progress — check with admin
  • Ensure historical engagements are indexed (background worker must be running)
  • Verify you have permission to access the historical data

"AI model not configured" error

  1. Navigate to Administration → AI Models
  2. Add at least one model (Ollama for on-prem, or Gemini/OpenAI for cloud)
  3. Test the connection
  4. Set the model as active

AI suggestion shows "Low Confidence"

  • This indicates the AI is uncertain about its output
  • Provide additional context and try again
  • Consider whether the task is too ambiguous for AI
  • Use the suggestion as a starting point but expect significant manual editing

Evidence summarization fails

  • Check the file format is supported
  • Ensure the file is not corrupted
  • Large files may hit token limits — try summarizing specific pages
  • Scanned PDFs require OCR which may be unavailable offline