New Data Connectors: Salesforce, HubSpot, and SAP

Enterprise data integration remains one of the most time-consuming challenges facing data teams. Connecting to business-critical systems like CRM and ERP platforms requires deep knowledge of proprietary APIs, complex authentication flows, rate limiting strategies, and schema mapping logic. Today we are releasing production-ready connectors for Salesforce, HubSpot, and SAP — three of the most requested integrations from our enterprise clients — dramatically reducing the time and expertise required to bring this data into the platform.

These connectors are not simple API wrappers. They are full-featured integration pipelines that handle the complete lifecycle of data extraction, from initial full loads through ongoing incremental synchronization. Each connector has been built in collaboration with clients who rely on these systems daily, ensuring that the implementation addresses real-world requirements rather than theoretical API coverage.

Connector Capabilities

Each connector supports full and incremental extraction modes, configurable scheduling, automatic retry with exponential backoff, and comprehensive error reporting. The Salesforce connector handles both standard and custom objects, including complex relationships like master-detail and lookup fields. The HubSpot connector covers contacts, companies, deals, tickets, and custom objects, with support for HubSpot's association model. The SAP connector integrates with both SAP S/4HANA and SAP ECC through RFC and OData interfaces.

All three connectors support change data capture, enabling near-real-time synchronization of changes from the source system. Rather than repeatedly extracting full datasets, the connectors detect and process only the records that have changed since the last sync. This approach reduces both the load on source systems and the processing time required to keep the data platform current.

Automated Schema Mapping

One of the most powerful features of these connectors is automated schema mapping. When a connector first connects to a source system, it introspects the schema — including custom fields, relationships, and data types — and automatically generates a mapping to the platform's internal data model. Data engineers can review and customize these mappings, but in most cases the automated mapping provides a production-ready starting point that eliminates hours of manual configuration.

Data connectors should be invisible infrastructure. When they work well, data teams forget they exist and focus on what matters — turning data into decisions. Our goal is to make every connector reliable enough to forget about.

Incremental Sync and Consistency

Maintaining data consistency during incremental sync is more challenging than it appears. Records can be created, updated, deleted, or undeleted in the source system, and the connector must handle all of these operations correctly. Our connectors implement a robust change tracking mechanism that detects all types of mutations and applies them in the correct order, ensuring that the platform's copy of the data always converges with the source of truth.

The connectors also handle the edge cases that trip up simpler implementations: API rate limits that interrupt mid-sync, network timeouts during large batch transfers, schema changes in the source system between sync runs, and timezone inconsistencies in timestamp fields. Each of these scenarios has been tested extensively and handled gracefully, with clear error messages and automatic recovery where possible.

Error Handling and Observability

Comprehensive error handling and observability are built into every connector. A dedicated monitoring dashboard shows sync status, record counts, error rates, and latency metrics for each configured connection. When errors occur, they are classified by severity and actionability — distinguishing between transient issues that will resolve on retry and configuration problems that require human attention. This classification helps data teams focus their attention where it is most needed.

Each connector also maintains a detailed sync log that records every API call, transformation, and load operation. This log serves both as an audit trail for compliance purposes and as a debugging tool for investigating data quality issues. When a downstream analyst questions a particular data value, the sync log provides a complete chain of custody from source system to platform table.