ClickHouse
ClickHouse Blog
ClickHouse shipped tooling and infrastructure upgrades aimed squarely at AI-native workloads, while the core database absorbed observability and data lake capabilities. The release of clickhousectl in beta positioned the CLI as a bridge between local development and cloud operations for both humans and AI agents, while version 26.3 made async inserts the default and introduced materialized CTEs. Simultaneously, ClickStack accumulated role-based access control, AI-powered log summarization, and embedded observability UI in the main ClickHouse binary, and the ecosystem expanded to native support for Iceberg and Delta Lake formats—reflecting a shift toward treating ClickHouse as the unified foundation for real-time analytics, observability, and agent-driven workflows at massive scale.
March built out ClickHouse's ecosystem and production-ready capabilities across analytics, observability, and AI workloads. Version 26.2 graduated text-index and QBit data types to production, embedded ClickStack directly into the binary for zero-setup observability, and shipped the async-native Python client with 1.16x better throughput under concurrency. The month also brought full-text search to GA, expanded cloud integration with unordered GCS ClickPipes and data lake support for Iceberg and Delta Lake, improved auto-scaling latency from 30 hours to 3 hours, and launched the official .NET client at 1.0.0 stable.