Setting Up Background Processing

Background Processing lets Claira run a bulk scan server-side. You start it the same way you start any bulk scan, but you do not need to keep Nuix Discover open while it runs — Claira finishes the work in the background and emails you when it's done. You can also kick off other background scans in parallel.

To enable Background Processing on a case, a case admin needs to do a one-time setup: create a dedicated Nuix Discover user for Claira, generate a personal API token for that user, and paste the token into Claira's settings. This page walks through each step.

Why a dedicated user?

When you run a bulk scan in the foreground today, Claira uses your Discover session to write coding back to the case. That session disappears the moment you close the browser tab — which is fine for short runs but not for an overnight job.

Background Processing keeps a long-lived Nuix Discover credential on Claira's side so it can write coding back hours or days after you click Start. To make that credential auditable, we recommend giving it to a dedicated user (for example, claira-service@your-firm.com) rather than tying it to a real reviewer. Every write-back from a background scan will then appear in Discover's audit trail as having been performed by that service account.

Step 1: Create a service user in Nuix Discover

1. From Portal Home, open Users.
2. Create a new user — for example claira-service@your-firm.com.
3. Give the user a strong password and store it somewhere safe (you will only need it once, to generate the API token in the next step).

Step 2: Add the service user to your case

1. Open Case Setup for the case where you want to enable Background Processing.
2. Click Users and add the service user you just created.
3. Give the service user enough permission to read documents and apply coding to your destination fields. A case-admin role is the simplest option; a coding-only role works too.

Step 3: Generate a personal API token

1. Sign in to Nuix Discover as the service user.
2. Open the user menu and follow your Discover deployment's instructions for generating a personal API token. (If your portal does not offer the option, ask your Nuix administrator to enable personal API tokens for your environment.)
3. Copy the token. It is a long string and will only be shown once — store it somewhere safe.

Step 4: Paste the token into Claira

1. Open Claira inside Discover on the case where you want to enable Background Processing.
2. Open Claira's settings and find the Background Processing section.
3. Paste the token into the Personal API token field and click Verify and save.
4. Claira verifies the token against Nuix and shows the service user's name once the save succeeds.

You're done. From now on, anyone with bulk-scan access in this case will see a Run as background processing task checkbox in the bulk-scan confirmation popup.

Running a background scan

1. Select your documents and start a bulk scan the way you usually do.
2. In the confirmation popup, check Run as background processing task.
3. Click Start bulk scan. The task appears immediately in Bulk Tasks with a Background badge and a status line such as "Queued at Vertex AI".
4. Close the tab if you want. When the task finishes, the initiating user receives an email with the result counts and a link back to Bulk Tasks in Claira.

Rotating or revoking the token

If you ever need to rotate the API token — for example, because the service user's password was changed — repeat Step 3 to generate a new token, then paste it into Claira's settings. Saving a new token automatically revokes the previous one.

To completely disconnect Background Processing on a case, click Revoke in the Background Processing settings panel. Existing background tasks already in flight will continue to completion, but no new background tasks can be started until you save a new token.

What gets written back

When the background scan finishes, Claira writes a summary of each verdict to the destination field you chose, the same way a foreground scan would — except the format is structured (responsiveness, confidence, rationale, citations) rather than free-form text. Reviewers can read the verdict directly in the case, or export the field and continue from there.