Level AI Data Export — Schema changes
Last updated: August 28, 2026
Effective date of change: August 25, 2026
Summary
Level AI has migrated its data export pipeline to a new CDC-based (Change Data Capture) mechanism. As part of this migration, the exported tables were fully refreshed, and new columns were added to them. The underlying data and its meaning are unchanged — what has changed is the schema of the exported tables.
There are three parts to this change:
Three HEVO* columns were removed which were always empty
New CDC metadata columns have been added to all exported tables
New business columns have been added to a set of specific tables
Apart from the three columns described in point 1, no columns have been removed. Your existing business columns remain present and populated.
If your team maintains ETL, transformations, or reporting on top of these exports, please review the sections below and validate your pipelines.
1. HEVO* to CDC field change
Note: This change is only applicable to customers who use file based exports
Your exports previously carried six HEVO* columns on every exported table. Only three of those ever contained data; the other three were always empty.
New CDC columns are added for the remaining three.
Previous HEVO* column | CDC field | Description |
__HEVO_MARKED_DELETED | _CDC_HARD_DELETED | Marker indicating the record was deleted in the source system |
__HEVO_LOADED_AT | _CDC_ICEBERG_TS_MS | Timestamp at which the record was loaded into intermediary layer |
__HEVO_INGESTED_AT | _CDC_SOURCE_TS_MS | Timestamp at which the change occurred in the source system |
__HEVO_XMIN | Removed | |
__HEVO_DATABASE_NAME | Removed | |
__HEVO_SCHEMA_NAME | Removed |
2. New business columns by table
The following columns have also been added to specific tables. The same list is available here: New columns by table
Table | New columns |
ACCOUNTS_USER | IS_BOT, LANGUAGE, PERSONA |
ACCOUNTS_USERORGANIZATIONS | DELETED |
AGENT_COACHING_COACHINGSESSION | CREATED_BY_ID, METADATA, PROJECT_ID, SOURCE |
JOURNEY_JOURNEY | IS_ARCHIVED, STATE |
LEVEL_ASR_ASRTRANSCRIPTIONLOG | POLARITY, PUB_SUB_LOG_ID, SESSION_AUDIO_LOG_ID |
LEVEL_NLP_CONCERNMININGHEADTOPIC | TOPIC_TYPE |
LEVEL_NLP_CONCERNMININGSUBTOPIC | TOPIC_TYPE |
LEVEL_NLP_JOURNEYCONCERNMININGOUTPUT | LAST_UPDATED_ASR_LOG_ID |
QA_CALIBRATIONS_CALIBRATIONSESSION | DISPUTE_STATUS, IS_REDACTION_ENABLED, USE_WEIGHTED_CALIBRATION_SCORE |
QUALITY_ASSESSMENT_CUSTOMFORM | GLOBAL_PROJECTS_ACCESS, SOURCE, TEMPLATE_USED, VERSION_NUMBER, VERSION_STATUS |
QUALITY_ASSESSMENT_CUSTOMFORMQUESTION | AUTO_QA_SETUP_TYPE |
QUALITY_ASSESSMENT_INSTASCOREREQUEST | COMPLETION_TIME |
QUALITY_ASSESSMENT_QAAUTOMATIONRULE | DEFAULT_RUBRIC_ID_ID, GOAL_PREDICTION_NOTIFICATION, LAST_PREDICTION_RUN_AT, MAX_RUN_COUNT, MODERATOR_SAMPLE, PREDICTION_PROGRESS, WORKLOAD_TIMEOFF |
What we ask you to check
Please confirm that your ETL, transformations, and reporting are parsing exports correctly for loads from August 25, 2026 onward. Data consumption pipelines that tolerate additional columns are generally unaffected.
Support
If you find an issue or need help interpreting any of the new columns, please reach out to your Technical Account Manager or the Level AI support team.