Troubleshooting
Other languages
Solutions for common issues you may encounter while using Claira for AI-powered document review.
Troubleshooting
This page covers the most common issues Claira users encounter and how to resolve them. If your problem is not listed here, or if these steps do not help, contact support with a description of the issue and any error codes you see.
Claira pane is blank or not loading
Why this happens: The Claira pane needs to be included in your Nuix Discover workspace layout, and Claira must be activated for your case and user group.
How to fix it:
- Check that your workspace layout includes the Claira pane. If you do not see it, ask your case administrator to add Claira to a workspace layout.
- Confirm that Claira has been activated for your case. Activation is case-level -- if the case has not been activated, the pane will not load content.
- Verify that your user group has access to Claira. Your administrator controls which groups can see the Claira pane.
- Try refreshing the page. Occasionally, a page reload resolves loading issues.
Media scan (image / audio / video) hangs or never returns a result
Why this happens: In Image, Audio, or Video mode, Claira first downloads the source file from Nuix Discover, then sends that file for analysis. Text mode can still work when this download step is slow, blocked, or returns a generic Content-Type. See Media scans for the file formats accepted by each mode.
How to fix it:
- Confirm two network steps in your browser DevTools. You should see the Nuix download request complete (HTTP 200) before the orchestrator
.../api/v1/orchestrate/chat/completionsrequest runs. If the download stays pending, the issue is Nuix/network/file size, not the AI step yet. - Wait for large files. Very large downloads can take several minutes; Claira stops waiting after a timeout and shows a clear download error.
- 403/404 on download: See your Nuix administrator for document access, branded redactions, download size limits, and UIX session token requirements (downloads must use the UIX session token, not a personal API token).
Scan says the document's file is too large
Why this happens: A scan that runs in your browser sends the document's file to Claira in a single request and reads files up to 23 MB. A background scan uploads the file in parts and reads files up to 50 MB. The message gives the file's size and the limit that applies. Nothing was sent to the AI, and nothing is charged. See File size limits.
How to fix it:
- For a file up to 50 MB, select the document in Nuix Discover, start a bulk scan and check Run as background processing task. The case needs Background Processing set up, and Claira only suggests this when it is.
- To scan the document's text instead, switch Scan as to Text. Claira only suggests this when the document has extracted text.
- When you scan many documents, leave Scan as on Auto: Auto reads a document's extracted text when its file is too large, and skips the document only when it has no text either.
- For a file over 50 MB with no extracted text, have the document processed for extracted text (OCR or transcription) in Nuix Discover, then scan it as Text.
Image scan says the file is not a supported format
Why this happens: An Image scan sends the document's own file to the AI, so that file has to be in a format Claira can read. The message names the file type Nuix Discover lists for the document. Discover records a file's real format when it is ingested, so a scanned document saved as .pdf can turn out to be another type of file. See Media scans for the formats Image scan reads, including TIFF and JPEG 2000 images.
How to fix it:
- Check the document's Document Type in Nuix Discover. If it is a format that holds text, such as a Word or email file, switch Scan as to Text. If it is audio or video, switch to that mode.
- If the message says the document has no extracted text either, a Text scan can't read it. Save the original file as a standard PDF, for example with Save As or Print to PDF in a PDF viewer, load that copy into the case, and run an Image scan on it.
- If the message says Claira couldn't prepare the document's image file, the TIFF or JPEG 2000 file is damaged, has more than 500 pages, or is too large once converted. Split or re-save the file and load it again.
- Still failing? Report the problem with the document ID.
Scan button is greyed out
Why this happens: Claira requires both a saved prompt and a connected field before it allows you to scan. If either is missing, the scan button remains disabled to prevent incomplete scans.
How to fix it:
- Go to the Prompt area and make sure you have a prompt written and saved.
- Go to the Settings area and confirm that a Nuix Discover field is connected.
- If both are in place and the button is still greyed out, check that the document viewer has a document loaded (for single scans) or that you have selected documents in the browse list (for bulk scans).
Results look wrong or inconsistent
Why this happens: In most cases, inconsistent results point to a prompt that could be more specific. The AI follows instructions literally, and small ambiguities in your prompt can lead to variable output across different documents.
How to fix it:
- Refine your prompt. Be more explicit about what you want. Define key terms, specify the exact output format, and list the allowed response values. See the Prompting Overview for detailed guidance.
- Test on known documents. Run your prompt on documents where you already know the correct answer. This helps you isolate whether the issue is with the prompt or the document.
- Compare scan settings and rerun. If results vary, rerun the same prompt on benchmark documents and compare outputs before and after each change.
- Verify the document has extracted text. If the source document has no text, garbled text, or very little text, switch the Scan as dropdown to Image, Audio, or Video to match the document's source file.
Bulk scan stopped or failed
Why this happens: Bulk scans can stop for several reasons, including browser interruptions, token exhaustion, or errors during processing.
How to fix it:
- Check whether the browser window was closed. Bulk scans require the browser window to stay open. If it was closed or the tab was navigated away, the task will have stopped.
- Check your token balance. If you ran out of tokens mid-task, the scan will stop. Verify your balance in the Usage popover or the bulk scan confirmation area.
- Look for error codes. If specific documents failed, Claira displays an error code for each one. Check the Error Codes page for resolution steps.
- Resume if available. A scan that stopped before finishing shows an amber Incomplete badge in the Bulk Scan Task History; open the task to see how many documents still need scanning, then click Resume Task to finish them (already-scanned documents are not re-scanned or re-charged). If you don't want to finish the run, Cancel task releases the reserved tokens. See If a scan stops early.
Cannot connect fields
Why this happens: Field connections require the field to exist in Nuix Discover and to be accessible from the Claira integration.
How to fix it:
- Verify the field exists in Nuix Discover. Open your case in Nuix Discover and confirm that the field you want to connect is present and has the correct type.
- Check the composite field ID. Claira uses a composite ID to identify fields. If the ID was entered manually, double-check it for typos.
- Confirm Claira is enabled for your case. If Claira has not been activated by an administrator, field connections will not be available.
- Check permissions. Your account needs write access to the target field. Ask your case administrator to verify your permissions.
Still stuck?
If none of the steps above resolve your issue:
- Note the exact error message or behavior you are seeing.
- Record the time, case name, and affected document IDs if applicable.
- Check the Error Codes page for specific error code guidance.
- Contact us with these details at support@claira.to. See Reporting a Problem for what to include.
Need help? Contact us at support@claira.to.
Was this page helpful?
Continue reading
Need more help?
Contact our support team at support@claira.to — we are here to help.