Goodbye to 'Email Me a File': Databricks OpenSharing Unifies Data and AI Agents

Databricks has launched OpenSharing, the evolution of its open-source Delta Sharing protocol, which now allows AI agents to directly access live data without the need to export files. This update eliminates the cumbersome 'email me a file' process and establishes a standard for sharing data securely and in real-time with artificial intelligence systems.

Impact for SysAdmins and DevOps

OpenSharing simplifies data integration across platforms, reducing manual workload. Infrastructure teams can now configure direct access to shared datasets, avoiding duplication and ensuring consistency. Additionally, the protocol supports Apache Iceberg, Parquet, and other formats, making it easy to adopt in existing environments.

For system administrators, this means fewer ETL scripts and fewer synchronization issues. AI agents can consume fresh data without human intervention, improving operational efficiency. However, it also introduces new security challenges, as discussed in our article on AI agents under attack.

Business Implications

From a business perspective, OpenSharing accelerates time-to-market for AI-based applications. Data teams can share datasets with partners and clients without friction, maintaining granular access controls. This is critical in industries like finance or healthcare, where data governance is key.

The ability of AI agents to access live data enables more accurate analysis and real-time decisions. For example, a sales agent could query updated inventories instantly, improving customer experience. For more on AI automation, check out our post on n8n and AI.

OpenSharing and Security

OpenSharing's architecture includes authentication and authorization based on standards like OAuth, allowing companies to share data without exposing their internal systems. However, the attack surface expands with the integration of AI agents. We recommend implementing secure VPNs and firewalls and conducting periodic penetration tests.

In summary, OpenSharing is a step forward toward data interoperability for AI, but its success will depend on secure implementation. As Acronis points out, AI and infrastructure must go hand in hand.

Source: The New Stack. ForgeNEX analysis.