Skip to Content
ChangelogLatest

Changelog

New features and improvements in Zeotap Agentic CDP. The changelog is updated with every release, and the changes released on the same day are listed together.

November 9, 2026

Improvements

  • Reliability improvements for BigQuery destinations.

November 6, 2026

New

  • a separate permission now controls who can sign a loader or destination in with an existing warehouse credential.
  • BigQuery syncs can use snapshot mode, which keeps each run’s full output in the table, stamped with its run.

Improvements

  • BigQuery destinations can use a Zeotap Agentic CDP-managed warehouse credential, writing as the same service account the source reads as, with no key to manage.
  • BigQuery and Snowflake syncs now fill a newly mapped column for rows already in the table.

November 5, 2026

New

  • BigQuery destinations can receive lists of objects, written as JSON text.
  • Training certificates. Pass the final assessment for your learning path to get a certificate you can print, save as a PDF, share as an image on LinkedIn and add to your LinkedIn profile.
  • Reuse warehouse credentials in loaders and destinations. Set up a BigQuery, GCS, Snowflake, Databricks, Redshift or ClickHouse loader or destination by choosing an existing warehouse credential instead of entering its key again.

Improvements

  • Steadier event forwarding at scale. Event forwarding keeps pace as you add more event sources.
  • Steadier realtime audiences. Realtime audiences handle growing event volumes more smoothly.
  • Changing a warehouse’s key. A key entered on a warehouse’s page now creates a new credential for that warehouse alone; change a shared credential’s key from its own page.
  • Reliability improvements for warehouse credentials.

November 4, 2026

New

  • Predictive models. Train propensity, churn and value models inside your own BigQuery, Snowflake or Databricks warehouse, and target audiences, syncs and journeys by score or percentile.
  • Bring your own scores. Use scores from a model trained elsewhere in audiences, syncs and journeys, kept current through one view per model.
  • RFM segmentation. Segment customers by recency, frequency and monetary value with a guided wizard.
  • Feature analysis. See which columns in a table are useful as model features before you build one.
  • Zeotap Agent in Gemini Enterprise. Buy Zeotap Agent on Google Cloud Marketplace and use it from your Gemini Enterprise app.
  • Billing accounts in organization settings. Register the Google Cloud billing accounts your company buys on, so your Marketplace purchases are recognised faster.
  • Choose exactly which WhatsApp templates to bring into your catalogue, instead of importing all of them at once — and remove any you no longer need.

Improvements

  • Folder actions a view-only folder doesn’t allow are now greyed out, with a note saying why.
  • The AI agent can now create, train, score and promote predictive models.
  • Refreshing your WhatsApp catalogue only updates the templates already there; it no longer pulls in new ones on its own.

November 3, 2026

New

  • Sites. Personalize your website from realtime audiences, realtime attributes and profile fields, choosing exactly what each page may read. Available on request.
  • Microsoft Fabric loader. Load tables and views from a Fabric Warehouse or Lakehouse SQL analytics endpoint, with full refresh, incremental or merge per table.

Improvements

  • One Mobile Ad ID box in Audience Boost. Select IDFA and GAID together with a single Mobile Ad ID option.
  • More dependable BigQuery loaders. BigQuery-to-BigQuery loaders run incremental and merge streams consistently, and their landed streams are ready to prepare in Data Prep.
  • Every column from file-based loaders. File-based loaders pick up columns that first appear in later rows or files, and Data Prep carries each one into your prepared tables.
  • More dependable loader runs. Retried loader reads resume from the last position loaded, and loaders created through the API get each stream’s default sync mode.

November 2, 2026

New

  • Braze arrays of objects. Sync a table with one row per object into an array-of-objects custom attribute on Braze profiles, kept in step by a mirror sync.

Improvements

  • More dependable Braze syncs. Braze syncs behave more predictably in every sync mode.
  • More predictable audience sync modes. Each sync mode delivers more consistently across destinations.
  • Multi-column primary keys for loaders. Key a merge-mode loader stream on more than one column by separating the columns with commas.
  • Exact rolling time windows on Snowflake. Rolling day and week windows in audience and trait filters stay exact across daylight-saving changes.

October 31, 2026

Improvements

  • New roles editor. Organization settings → Roles now edits one role at a time, grouped by product area, with search, grant all or clear all per area, and a side-by-side compare with another role.
  • Edit single role permissions through the API. A new endpoint grants or revokes individual permissions on a custom role without touching the rest.
  • Shareable Organization settings links. A reload or a copied link opens Organization settings on the organization you were editing.

October 30, 2026

Improvements

  • Hours and calendar units in event filter dates. Relative-date operators in event forwarding filters now take hours and calendar days, weeks, months and years.
  • Relationship graph keeps your layout. Adding a model places it next to the model it relates to, without rearranging the cards you have placed.

October 29, 2026

New

  • PII ID columns. Mark a column such as a hashed ID, cookie or device ID as a PII ID to keep it out of audience filters and value suggestions while it stays available to syncs.
  • Meta Custom Audiences match on your external ID. Map your own user ID, the one you send with the Meta Pixel or Conversions API, as the External ID.
  • Relative-date operators in event forwarding filters. Filter timestamp fields with within, not within, between and not between, using rolling or absolute dates.

Improvements

  • One mobile ad ID column for Audience Boost. Map a column that holds both IDFAs and GAIDs once, as Mobile Ad ID.
  • BigQuery loaders stage files in your own project. Load files are now staged in your BigQuery project by default.
  • Add several people to a group at once. Tick everyone in a group’s member picker and add them in one step.
  • Smarter Audience Boost auto-detect. Auto-detect maps a mobile ad ID column as Mobile Ad ID.

October 28, 2026

New

  • Google Ads (Partner) destination. Upload Customer Match audiences to Google Ads by entering a customer ID and linking our data partner account, with no Google sign-in or developer token.

Improvements

  • Google Sheets asks only for the spreadsheets you pick. Choose a spreadsheet with the Google Drive picker, and the connection gets access to that file only.
  • More dependable golden records. Identity resolution builds profiles only from events that carry an identifier.
  • Google Data Manager and Google Sheets loader improvements. Data Manager handles removals and addresses more dependably, and the Sheets loader loads tabs with numbers, checkboxes and dates.
  • Smoother field mappings and computed attributes. Filters on related columns stay visible alongside transforms, relative date filters work on related columns and computed attributes, and syncs Zeotap Agent starts use your usual destination settings.

Earlier updates

Last updated on