> ## Documentation Index
> Fetch the complete documentation index at: https://docs.presenton.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Presentation generation problems

> Fix generation that will not start, takes too long, fails, or produces an unexpected presentation.

## Generation does not start

1. Make sure the prompt explains the presentation topic and that every required option has been selected.
2. Wait for attached files to finish uploading or processing before selecting **Generate**.
3. Refresh Presenton, start a new presentation, and retry with a short prompt and a small slide count.
4. If that works, add the original files and settings back one at a time to find what causes the problem.

## Generation takes too long or stops

* Do not repeatedly select **Generate**; this can start duplicate work.
* Try fewer slides and a smaller source file.
* If one file is very large, include only the pages or sheets needed for the presentation.
* Note whether the process stops during file processing, outline creation, slide generation, image selection, or final rendering. This detail helps support locate the failure.

## The presentation was created, but the result is not right

* State the audience, purpose, tone, and desired outcome directly in the prompt.
* Review the outline before generating slides. Correct missing topics, ordering, or emphasis there first.
* Ask for exact facts to be preserved and avoid asking Presenton to infer figures that are not in the source.
* Use a source with clear headings and readable text when the presentation misses important details.
* Regenerate only after changing the prompt, outline, source, or settings; retrying the same input may produce a similar result.

## A file-based presentation fails

Try the same prompt without the file. If it works, open the source file on your computer, confirm it is not password protected, and upload a smaller copy. See [Files, content, and images](/archive/v0.9.4-beta/troubleshooting/content-and-media) for format and extraction checks.

## If you manage a self-hosted instance

* Verify the selected text provider URL, model name, credentials, network access, and account quota.
* Confirm the model meets the structured-output and context requirements for the installed Presenton release.
* Check application logs for provider, schema, rate-limit, timeout, rendering, storage, or memory errors.
* Confirm `/app_data` is writable, has sufficient free space, and persists across container restarts.

If generation fails for every user with a small known-good prompt, the problem is likely deployment-wide. If it fails for only one prompt, presentation, or file, preserve a sanitized copy of that input for testing.