Batch-generate AI fields across hundreds of records

After bulk import or when older records never received AI values, workspace admins run background jobs that fill AI field (Long text) and AI field (Score) columns in one pass — filtered by date and scoped to missing, failed, or all records.

500

Records per batch

Earliest by created date when more match

BG

Background jobs

Leave the page; progress persists

2

AI field types

Long text summary and Score 0–100

Backfill AI fields at scale

Batch jobs in Setup after import or schema changes.

AI fields shine in Operations when every record has a summary and score. Manual sparkle refresh works for one-offs; batch jobs handle the backlog after migration or schema changes.

Setup → AI and Agents is built for controlled mass updates. Preview counts before you start, leave the page while jobs run, and review recent job history when batches finish.

  • Background batch jobs

    Queue up to 500 records per batch. Jobs continue after you navigate away; cancel stops after the current chunk.

  • Date and state filters

    Narrow by created date range. Process missing or failed values only, or regenerate all AI fields on matching records.

  • Pairs with import

    Bulk import skips AI generation by design. Run a batch here after load when summaries and scores should catch up.

When

When batch generation beats one-by-one refresh

Use batch jobs after migration, backfill, or when AI fields were added to an existing object.

  • Only objects with AI fields appear in the object dropdown
  • One batch job runs at a time per workspace
  • Individual sparkle refresh in Operations still works alongside batches

Filters

Filters that keep runs predictable

Object, created-from, created-to, and processing mode shape the preview count before you commit.

  • Created date range is inclusive on created_at
  • Missing or failed only skips records with ready AI values
  • Regenerate all recalculates even when values already exist

Limits

500-record cap and queue behavior

Large match sets are trimmed to the earliest 500 by created date so jobs stay bounded.

  • Preview count updates when filters or mode change
  • Records currently generating are skipped until they finish
  • Each record refreshes all AI fields on that object in one pass

Progress

During and after a batch

Progress, cancel, and job history live on the same Setup page.

  • Start batch moves the job to running with a progress bar
  • Cancel preserves values already written in finished chunks
  • Recent jobs list succeeded, failed, and status per batch

Access

Governance and permissions

Batch AI generation is a workspace admin capability tied to Setup access.

  • Members still consume AI values in Operations and Live Views
  • AI field types are defined in Setup → Fields
  • Permission profiles control who sees AI columns on records

Short answers before you request access.

Does bulk import auto-generate AI fields?
No. Import writes mapped columns only. After import, run a batch in Setup → AI and Agents when you want AI summaries and scores populated across the loaded records.
What happens if more than 500 records match?
Only the earliest 500 by created date are queued. Narrow the date range or run multiple batches to cover the rest.
Can I regenerate AI values that already exist?
Yes. Choose Regenerate all to recalculate every AI field on matching records. Choose Missing or failed only to skip records that already have ready values.
Do batch jobs block Operations?
No. Jobs run in the background. You can leave the page, keep working in Operations, and return to check progress or recent job history.

Request access

Tell us about your team. We'll reach out when your workspace is ready.

AI and Agents — Batch AI Field Generation | OSTRATA