Docs / Troubleshooting / Contract Issues

Contract Issues

Troubleshoot contract generation and editing problems.

Contract generation relies on your connected AI provider. Most issues stem from provider errors, insufficient input, or file format limitations. Below are the most common problems and how to resolve them.

Generation fails

If contract generation starts but fails with an error or never completes:

  • Check your AI provider. Generation requires a working AI connection. Go to Settings > AI Provider and verify your key is valid and the provider is reachable. See AI Not Responding for detailed troubleshooting.
  • Insufficient input data. QuickContract needs at minimum a contract type and the parties involved. If you are generating from a meeting, make sure the meeting has a completed transcript — generation cannot proceed from an empty or failed transcription.
  • Try again. Transient provider errors (timeouts, 500 errors) are common. Wait a few seconds and click Generate again. If the failure is consistent, switch to a different AI model in Settings.
Ollama models may struggle with long contracts

Local models running via Ollama, especially smaller ones (7B parameters), may not have enough context window to generate complete contracts. If generation fails or produces garbled output with Ollama, try a cloud provider or a larger local model.

Contract is incomplete

If the generated contract is cut off or missing sections:

  • Token limit reached. AI models have a maximum output length. Very long contracts (e.g., 20+ pages) may exceed this limit. QuickContract automatically truncates input to stay within model limits, but especially complex contracts can still hit the output ceiling.
  • Use a model with a larger context window. Models like Claude 3.5 Sonnet, GPT-4o, and Gemini 1.5 Pro support large context windows and longer outputs. If your current model produces truncated results, switch to one of these in Settings > AI Provider.
  • Generate in sections. For very large contracts, consider generating the core agreement first, then using QuickEdit to add supplementary sections, schedules, or appendices one at a time.

Formatting issues

If the contract text looks correct but the formatting is off — missing headings, broken numbering, or inconsistent spacing:

  • Use the contract editor. QuickContract's editor renders the AI output as formatted text. Minor formatting irregularities from the AI can be fixed directly in the editor by adjusting headings, lists, and spacing.
  • Export and reformat. If the formatting is significantly off, export the contract to DOCX or PDF and make final adjustments in a word processor. The export preserves headings, numbered clauses, and basic styling.
  • Different models format differently. Each AI model has its own formatting tendencies. If one model consistently produces poorly formatted output, try switching to another. Claude and GPT-4 models tend to produce clean, well-structured legal documents.

Upload not working

If you are trying to upload an existing contract and it fails or the extracted text is garbled:

  • Supported formats: QuickContract supports .pdf, .docx, and .txt files. Other formats (e.g., .doc, .rtf, .pages) are not currently supported.
  • Scanned PDFs: If your PDF is a scanned image rather than selectable text, QuickContract cannot extract the content. You will need to run the document through an OCR tool first and then upload the resulting text-based PDF.
  • File too large: Very large files (over 10 MB) may take a long time to process or fail. If possible, reduce the file size or split the document into smaller parts before uploading.
  • Password-protected files: QuickContract cannot open password-protected or encrypted PDFs. Remove the password protection before uploading.
Tip

After uploading a contract, review the extracted text in the editor to check for any extraction errors before running AI analysis or comparison. Minor cleanup at this stage saves time later.