Daily Briefing — Friday, November 6, 2026
Vendors are racing to own the agent context layer — streaming serving tiers, semantic ontologies and open-catalog federation — while lineage for agent access remains the unsolved governance gap.
⇣ Jump To
Click any section or topic below to jump to it.
⚡ Quick Takes
| Story | Signal |
|---|---|
| ↗ Snowflake Introduces Autonomous Data Engineering: CoCo Cloud Agents, Datastream and Streaming Transformations | Snowflake pushes ingestion and agentic pipeline repair into the warehouse |
| ↗ What's New in Q4'26 Confluent Data Streaming Platform | Confluent adds a serving layer and agent context engine on top of Kafka |
| ↗ SAP promises to open out data flows in both directions | SAP BDC goes Iceberg and Polaris, with zero-copy sharing to the big platforms |
| ↗ Google Agentic Data Cloud gets cross-cloud Lakehouse | Google makes cross-cloud Iceberg queries GA with catalog federation |
| ↗ Snowflake expands agentic AI platform with business context and governance tools | Snowflake extends semantic views into a governed business ontology for agents |
| ↗ Securing the agentic enterprise with data lineage | Sponsored piece frames lineage as the security perimeter for agent workflows |
Snowflake — November 2026
Snowflake pitched self-healing pipelines: CoCo Cloud Agents and Automations that diagnose and repair failures with engineers approving production changes, alongside Snowpipe Streaming, a Kafka-compatible Snowflake Datastream, Streaming Transformations and Interactive Analytics. The release cites Snowpipe Streaming at up to 20 GB/s per table with ~3-second median readiness. GA vs. preview status for several pieces is not clear from the release, so treat it as direction until docs confirm. Snowflake moves further into Kafka-compatible ingestion and in-platform streaming transforms, narrowing the gap with Confluent-plus-warehouse stacks.
✍️ Snowflake Inc. (via Business Wire) · Read article →
Confluent — November 2026
Confluent adds Query as a fifth platform pillar, centered on Lightning Tables, a managed read-only serving layer over Kafka data. Managed Model Inference (Bedrock, Vertex AI, Azure OpenAI), the Real-Time Context Engine serving agents over MCP, and managed MCP servers are GA; on-prem Flink AI functions and an NL2SQL assistant are open preview. Tableflow extends to Google Cloud and BigQuery. Kafka vendors are absorbing the serving and agent-context tier, which changes the build-vs-buy line for real-time feature and context stores.
✍️ Confluent · Read article →
SAP — November 2026
At TechEd, SAP added native Apache Iceberg support to Business Data Cloud plus a universal catalog based on Apache Polaris, and says BDC Connect enables zero-copy sharing with Databricks, Snowflake, Google, AWS and Microsoft. Dremio's engine, acquired in May, lets BDC users reach SQL Server, MongoDB, Oracle and PostgreSQL without moving data. The article gives no GA dates. The system of record for much of enterprise finance and supply chain is joining the open-catalog ecosystem, which makes Iceberg REST federation the default integration surface.
✍️ Lindsay Clark, The Register · Read article →
Techzine — November 2026
Google's Iceberg-based Lakehouse now queries AWS and Azure data over private interconnect without variable egress charges; cross-cloud interconnect and caching are GA. Catalog federation covers Databricks Unity, Snowflake Horizon and AWS Glue, and AlloyDB can query Iceberg tables directly. Performance and cost claims (BigQuery 40% faster, 50% cheaper per query) are Google's own. Cross-cloud Iceberg federation without egress penalties weakens the data-gravity argument for consolidating on a single cloud.
✍️ Berry Zwets, Techzine Global · Read article →
SiliconANGLE — November 2026
Snowflake announced Business Ontology, an extension of semantic views holding approved business definitions, now in private preview. Cortex AI Gateway gained abuse detection, tracing, sandboxes, AI cost governance and dynamic model routing, and Cortex Training entered public preview for post-training open-weight models. Snowflake also expanded its Oracle partnership to offer its platform on OCI. Semantic definitions are becoming the governed grounding layer for agents, and gateway-level cost and routing controls make that layer operable.
✍️ Kyt Dotson, SiliconANGLE · Read article →
MIT Technology Review Insights — November 2026
Sponsored content arguing that lineage is the new security perimeter once agents hand data to other agents with no human checkpoint. It offers a scenario rather than data, so read it as vendor framing. The practical point stands: lineage capture has to extend across agent-to-agent handoffs, not just pipelines. Agent access is the gap in current lineage standards, which makes metadata capture at the data-movement layer a design requirement.
✍️ MIT Technology Review Insights (sponsored by Cyberhaven) · Read article →