Generation Issues

Troubleshooting incomplete generations, retries, and recovery

My Scribe Generation Stopped

If your scribe generation stops before completing all sections, this is called a partial generation. This can happen due to:

  • AI model timeouts or temporary outages
  • Network interruptions
  • Token limit reached mid-generation
  • AI model rate limits

When a scribe is partially generated, it is saved with a partial status — no content is lost. The system will automatically attempt to retry failed sections with exponential backoff before giving up.

What to Do

  1. Check the progress indicator — If it shows a specific percentage, some sections were successfully generated
  2. Look for the "Complete Scribe" button — This appears when a scribe has missing or failed sections
  3. Click "Complete Scribe" — The system will generate only the missing sections, picking up where it left off

Retrying Failed Sections

If specific sections failed:

  1. Open your scribe
  2. Sections that failed to generate will be marked with an error state
  3. Use the retry option to regenerate just those sections
  4. The system uses the same model and settings as the original generation

Generation Is Taking Too Long

AI generation speed depends on several factors:

  • Generation mode — Lightning is fastest (all sections in parallel). Lightning Medium and Pour Over are middle-ground options. Pour Over Slow Brew is slowest but most coherent.
  • Model choice — Premium models may take longer per section
  • Scribe length — More chapters and sections take more time
  • Server load — During peak times, generation may be slower

What to do:

  • Wait a few minutes — Lightning mode can finish a 24-chapter scribe in under 8 minutes; other modes take progressively longer
  • Don't refresh the page — progress tracking will resume if you navigate away and come back
  • If a scribe is stuck for more than 10 minutes with no progress, it will be detected as stale and you'll see a prompt to retry or complete it

"My Scribe Shows 100% But Sections Are Missing"

This is rare but can happen if the progress tracker and actual generation get out of sync. Use the "Complete Scribe" button to fill in any missing sections.

Content Quality Issues

If the generated content doesn't meet your expectations:

  • Try a different AI model — Premium models generally produce higher-quality output
  • Be more specific in your instructions — Detailed prompts produce better results
  • Use the Workspace — Rewrite individual sections with custom prompts
  • Enable web search — For factual topics, web-researched content is more accurate

Token-Related Generation Failures

If generation stops because you ran out of tokens:

  • Your partially generated scribe is saved
  • Purchase top-up credits or wait for your monthly reset
  • Use "Complete Scribe" to finish the remaining sections