Integrate.io logo

Integrate.io Release Notes & Product Updates

An overview of the latest updates and product improvements. Got a product update or feature you'd like to see supported? Let us know - support@integrate.io

October 2026: Sync your packages to GitHub, and promote them to Prod by pull request

ETL & Reverse ETL

October's release cycle gave teams more control over how packages reach production and clearer sight of what each run actually did. New connectors on both ends of the pipeline and steady connector work rounded out the month.

Sync Packages to GitHub and Promote Them by Pull Request

Packages have lived only inside the platform, so keeping their history in Git meant exporting JSON by hand and remembering to do it after every change. The GitHub integration syncs your packages and workflows to a repository you choose, so each saved version is tracked in Git next to the rest of your code. Teams running separate Dev, QA, and Prod accounts can go a step further with package promotion, which carries changes from one account to the next through reviewed pull requests, and Prod changes only after someone approves. We are rolling out this feature gradually, and if you want it now, let our team know. The documentation covers the integration, and package promotion has its own page.

Job Insights Shows Where Every Run Spends Its Time

Working out why a job ran long, or where its row counts went wrong, used to mean reading through logs and guessing from the totals at the end. Job Insights lays out each completed dataflow run as a timeline, with record counts for every component. A slow step or a source that read fewer rows than expected stands out at a glance, so tuning a package or answering a data question starts from evidence instead of a log search. See the documentation for what each part of the view reports.

Filters Put Advanced Search One Click Away in Every List View

Search bar modifiers have long let you narrow a list of jobs, packages, or connections, but only if you knew the syntax by heart. A Filters button beside the search bar puts those same options on screen in seven list views, from Jobs and Packages to Connections and Clusters. Anyone on the team can pull up failed runs on one package, or packages changed this month, without learning a query language first. Each choice lands in the search bar as a modifier, so typed searches keep working as before. The documentation lists the filters each view offers.

New Connectors

  • Outlook Source and Destination: The Outlook source reads calendar events and mail from a signed-in or shared mailbox, and the Outlook Calendar destination creates and updates events from pipeline records. Dates from a CRM or database land on a shared calendar and stay in step with their source instead of being copied over by hand (source and destination documentation).
  • NetSuite REST Destination: The NetSuite REST destination completes the pair with the existing NetSuite REST source and writes standard and custom records, transactions with line items included, by insert, update, or upsert. Upserts match on external ID, so a load can run again after a failure instead of leaving duplicate records to clean up (documentation).
  • ServiceNow Destination: The ServiceNow destination completes the pair with the existing ServiceNow source and writes records into an import set table or straight into a table over the same connection. Import sets run through your instance's transform maps, so records land under the rules your ServiceNow team already set instead of a parallel load path (documentation).
  • Oracle Fusion Source: The Oracle Fusion source reads objects across Financials, Procurement, CRM, HCM, and more from payables and receivables invoices to suppliers, purchase orders, customer accounts, and workers, all with incremental loads. Fusion data reaches the warehouse on a schedule instead of through report exports saved as CSV files (documentation).
  • Square Source: The Square source reads 14 objects, including payments, refunds, payouts, customers, and catalog items, with incremental loads on payments and refunds. Sales and customer data from Square joins the rest of your reporting on a schedule instead of through a manual export (documentation).
  • Fundraise Up Source: The Fundraise Up source reads nine objects, including donations, recurring plans, supporters, and the event log, with incremental loads on six of them. Donation and supporter data sits in the same warehouse as your CRM and finance data instead of in a separate export (documentation).
  • OneTouch Health Destination: Following September's OneTouch Health source, the destination creates and updates carer records through the OneTouch API over the same connection. Carer records maintained in another system reach OneTouch from a pipeline instead of being re-entered by hand (documentation).

New Features

  • ZIP Extraction in File Mover: File Mover can unpack ZIP archives at the destination and deliver the extracted files, into one folder per archive or all into one folder. Partner drops that arrive zipped reach their folder ready to load instead of waiting on a separate unzip step (documentation).
  • Excel Output on File Storage Destinations: SFTP, Amazon S3, Google Cloud Storage, and Azure Blob Storage destinations write Excel (.xlsx) files. Teams that expect a spreadsheet get one straight from the job instead of converting a CSV by hand.

Improvements

  • File Storage destinations default to intermediate storage, with a copy into an existing directory as the destination action.
  • Downstream Field Propagation shows a compact summary of fields not carried into child components, and newly added source fields appear in full until the next refresh.
  • Source and destination editors collapse Advanced settings by default, with a header that shows which options are switched on, and every component shares one settings layout.
  • File sources replace three header checkboxes with a single Header choice and add Semicolon and Pipe as delimiters.
  • Import and Export JSON is available to custom roles on accounts that have Source Control enabled.
  • Helm builds Salesforce to Snowflake pipelines, including packages that move several Salesforce objects at once.
  • The MCP server adds a save_package_version tool, so an agent can save a package's draft as a new version.
  • The MCP get_run_summary tool returns the records each source read and each destination wrote, per flow and in total, in one call (documentation).
  • Salesforce connections add Client Credentials as a connection option, so a pipeline can connect without signing in as a named user.
  • Salesforce destinations can write their success and error files to SharePoint.
  • The Snowflake destination supports Query Tags, so you can find the queries a job ran in Snowflake's query history.
  • The REST API destination accepts fields and variables in Custom Input mode, so you can build any JSON body from pipeline data.
  • File sources add a file_modified_at metadata field holding each file's last modification time in the source file system.
  • Two package variables tune SFTP writes: _FS_SFTP_WRITE_CHUNK_SIZE sets the chunk size, and _FS_SFTP_WRITE_MAX_UNCONFIRMED set to 0 suits servers that fail on several writes at once.
  • The SharePoint source reads files stored in Microsoft Teams as well as SharePoint sites.
  • The HubSpot source reads associations between objects, such as the companies and deals linked to a contact.
  • The Facebook Ads source retries temporary Graph API errors for up to 30 minutes, slows down near Facebook's rate limits, and re-creates reports Facebook marks as failed.
  • The Google Ads source runs on API versions v23 to v25, and v20 and v21 are removed after Google deprecated them.
  • The Google Ads source adds the Lead Form Submission Data report, so lead form entries load alongside your campaign data.
  • The Bing Ads source runs on a new connector that authenticates with a developer token.
  • The REST API source pages through DocuSign results, so full result sets arrive instead of only the first page.

Fixes

  • Incremental loads that use a variable as the reference date send the variable's value, so jobs no longer fail or skip the date filter.
  • _PACKAGE_LAST_SUCCESSFUL_JOB_SUBMISSION_TIMESTAMP, _WORKFLOW_PACKAGE_LAST_SUCCESSFUL_JOB_SUBMISSION_TIMESTAMP and their S3-escaped versions no longer skip data when a workflow child starts late.
  • Scheduler auto-retry works on schedules that run every 5 minutes or less, and auto-retry schedules no longer drop runs silently.
  • File Mover deletes source files only after they are confirmed delivered, and keeps them when any step fails or a file changes mid-job.
  • File Mover moves from Azure delete their source files again after delivery.
  • Long-running Salesforce source jobs finish instead of failing on a timeout error.
  • Concurrent jobs writing to the same file storage folder no longer collide over shared staging files.
  • Jobs that clean up intermediate storage on Amazon S3, such as Snowflake loads, fail quickly on a timeout.
  • Workflow variables that use AddToDate and other newer expression functions parse at job start.
  • The REST API destination accepts special characters in payloads and settings and sends single-field values, so writing rows to Google Sheets works.
  • The Salesforce API destination works on accounts that route traffic through the Integrate.io egress proxy.
  • Redshift connections that drop during connection setup retry automatically instead of failing the job.
  • The GA4 source retries temporary API errors on every call instead of failing the job.
  • API keys stay valid when someone signs in through SSO for the first time.
  • CSV imports work with Detect header row automatically when a source path field is also in use.

Tell us how package promotion fits your release process, and where Job Insights still leaves a support question unanswered.

10 days ago