One of the persistent limitations of large language models in enterprise settings has been the gap between what the model knows and what is actually happening in the business right now. A model trained on data through a certain date, even one with a 200,000-token context window, cannot tell you what your pipeline looked like yesterday or which customers are at risk of churn this month, unless you explicitly feed it that data. Anthropic's new real-time data connectors for Claude Enterprise are designed to close that gap.
The feature, announced April 20, 2026, allows enterprise administrators to configure Claude with read-only access to authorized data sources. When a user asks a question that requires current data, such as "What are our top five deals by value this quarter?" or "Which support tickets have been open for more than 72 hours?", Claude will automatically query the relevant connector, retrieve the data, and incorporate it into the response. No manual copy-paste of reports, no context-filling CSV uploads. The data pipeline is handled transparently in the background.
Supported Integrations
The initial connector library covers 14 data sources, spanning the most common enterprise data infrastructure:
- Salesforce CRM (Opportunities, Accounts, Contacts, Cases)
- Snowflake Data Cloud (read-only SQL queries)
- PostgreSQL and MySQL databases
- Google BigQuery
- HubSpot (deals, contacts, marketing data)
- Jira (issues, sprints, project tracking)
- Confluence (knowledge base and documentation)
- Slack (with explicit user consent and message scoping)
- ServiceNow (incidents, changes, CMDB)
- Workday (HR data, with strict role-based access controls)
- Microsoft SharePoint and OneDrive
- REST API connector (configurable for custom internal APIs)
- GraphQL API connector
- Amazon S3 (document and data retrieval)
"Within the first week, our revenue operations team asked 340 questions that would have previously required a data analyst to pull a report. Claude answered all of them in real time. The productivity impact is not incremental — it is transformational." — Head of Revenue Operations, enterprise SaaS company (beta customer)
Early Results
During the six-week closed beta, 23 enterprise customers tested the connectors across a range of use cases. The most common workflows were sales pipeline analysis (Salesforce), support queue management (ServiceNow and Jira), and data exploration (Snowflake and BigQuery). Customer feedback was overwhelmingly positive on utility, but the most significant finding was around time savings: tasks that previously required a data analyst and a 24–48 hour turnaround were being completed in minutes by non-technical business users querying Claude directly.
Security and data governance were the most common concerns raised during beta, and Anthropic has addressed them with a layered permission model. Connector access is configured by administrators and scoped to specific objects and fields. Role-based access controls ensure Claude can only retrieve data that the querying user is authorized to see. All connector queries are logged for audit purposes, and data retrieved via connectors is not retained or used in model training. For regulated industries, this last point is particularly important.
The real-time data connectors are available now for all Claude Enterprise customers at no additional charge beyond the existing enterprise subscription. The REST and GraphQL connectors, which require custom configuration, come with dedicated onboarding support. Anthropic has signaled that the connector library will expand to include Zendesk, Oracle, and SAP S/4HANA in the next product quarter.
