January 2026 Summaries

DoControl, a SaaS security company, transitioned from Postgres to ClickHouse Cloud to enhance real-time analytics capabilities for managing file access, permissions, and third-party app integrations, especially after onboarding enterprise clients with vast data volumes. Originally limited by Postgres's inability to handle high-volume, complex queries, the company adopted ClickHouse Cloud, enabling faster data ingestion and real-time analytics across millions of assets. This shift supports their AI-powered assistant, Dot, which utilizes ClickHouse MCP to process natural-language queries, offering clients quick insights into potential security risks. As DoControl scales, they focus on expanding AI functionalities, improving infrastructure, and maintaining real-time analytics capabilities, with ClickHouse at the core of these efforts.

Lovable, a rapidly growing Swedish startup, utilizes ClickHouse Cloud to manage observability and AI-powered debugging for its AI-generated applications and large language model (LLM) workflows. ClickHouse MCP allows engineers to efficiently analyze billions of logs using natural-language prompts, significantly reducing debugging time and maintaining high development velocity. The platform supports real-time analytics for millions of deployed applications, enabling fast query responses with a simple setup. Lovable's mission to democratize software creation has led to impressive growth, with over $100 million in subscription revenue within eight months of launching and more than 8 million users creating thousands of projects daily. The company leverages ClickHouse for scalable analytics, AI-driven debugging, and observability, enhancing both internal operations and user experience by offering real-time insights into app performance and security. With ClickHouse's flexibility, Lovable empowers engineers and users to explore data and understand system behavior without deep database expertise, thereby facilitating swift development and problem resolution.

ClickHouse has introduced the Official ClickHouse Kubernetes Operator to streamline the deployment and management of Open Source ClickHouse on Kubernetes, addressing the challenges of running a stateful, high-performance database in cloud-native environments. The Operator, available under the Apache-2.0 license, automates cluster provisioning, supports ClickHouse Keeper, enables easy scaling, and facilitates seamless upgrades while maintaining service continuity. Key design choices include using ClickHouse Keeper for coordination, adopting DatabaseReplicated as the default engine, implementing a StatefulSet per replica for detailed control, and ensuring TLS/SSL encryption for secure communication. The Operator aims to keep the setup simple by leveraging ClickHouse's existing capabilities, providing a thin operational layer. Users can quickly get started with ClickHouse Cloud and a $300 free credit offer, further supported by community engagement through contributions and feedback via GitHub, Slack, and other channels.

The integration between ClickStack and Temporal Cloud's OpenMetrics endpoint introduces high-performance observability to Temporal, a durable execution platform that simplifies error handling and state management in applications. Temporal ensures seamless execution of workflows, even amidst failures, by automatically recovering and resuming tasks. ClickStack, built on ClickHouse, offers scalable storage and querying of logs, metrics, and traces, making it ideal for analyzing large volumes of high-cardinality data. This partnership provides a unified observability solution, allowing users to track metrics efficiently and optimize workflow performance. The integration leverages the OpenTelemetry Collector's Prometheus receiver to gather metrics from Temporal Cloud, enabling teams to visualize and analyze system health through ClickStack's HyperDX UI. By combining Temporal's reliable execution capabilities with ClickStack's robust observability features, users can effectively monitor and enhance their mission-critical workflows.

The BigQuery connector for ClickPipes, now in Private Preview, streamlines the process of loading data from BigQuery into ClickHouse Cloud, allowing for fast exploration and prototyping by automating data migration tasks such as schema mapping and type conversion. This integration addresses challenges like the manual effort previously required for exporting data and configuring pipelines, offering efficient bulk data loads and fine-grained control over datasets and tables. Built on the PeerDB platform, it supports continuous ingestion and is designed to enhance the performance of interactive queries by functioning as a speed-layer atop BigQuery, with potential future capabilities for incremental data syncs. The connector utilizes BigQuery's native export jobs to manage data transfers cost-effectively and requires specific setup in Google Cloud, including a dedicated staging bucket. As the service evolves, its development is guided by customer feedback, with opportunities for design partnerships to refine features like incremental syncs, which remain a priority due to customer demand.

Buildkite transitioned to using ClickHouse Cloud for real-time test analytics, significantly enhancing its capabilities and reducing infrastructure costs despite a substantial increase in data ingestion. This shift enabled Buildkite to retire several systems, including DynamoDB and reduce reliance on others like Aurora and Flink, while managing a massive influx of test events at a peak rate of 25,000 per second. ClickHouse's speed, ease of use, and SQL compatibility allowed Buildkite to offer customers more flexible and intuitive analytics, transforming how they interact with test data by supporting on-demand queries. This change not only simplified Buildkite's tech stack but also improved performance and reduced costs, leading to savings and an enhanced customer experience. The adoption of ClickHouse Cloud has empowered Buildkite to deliver more robust and rapid analytics, meeting customer demands for real-time, customizable data insights, and allowing the engineering team to focus on innovation rather than extensive database management.

ClickHouse, a leader in real-time analytics, data warehousing, observability, and AI/ML, announced its acquisition of Langfuse to enter the LLM observability market and introduced a native Postgres service to integrate transactional and analytical workloads. The company completed a Series D funding round, securing $400 million led by Dragoneer Investment Group, with participation from notable investors, and reported over 250% year-over-year ARR growth. ClickHouse Cloud supports over 3,000 customers, with recent expansions including Capital One and Airwallex, and maintains a strong customer base with major brands like Sony and Tesla. CEO Aaron Katz highlighted the platform's performance and cost-efficiency for demanding data workloads, enabling developers to build AI-driven applications on a robust infrastructure. Dragoneer emphasized ClickHouse's role in supporting large-scale AI systems with performance and reliability. Langfuse, focusing on LLM observability, has rapidly gained adoption and will enhance ClickHouse's offerings. Additionally, ClickHouse's new Postgres service, developed with Ubicloud, allows seamless integration of transactional and analytical processes for modern AI applications, underscoring ClickHouse's expanding global ecosystem and product growth.

ClickHouse has achieved the AWS AI Competency in the Software partner track, marking its recognition as a provider of enterprise-grade autonomous AI solutions within the newly expanded agentic categories by AWS. These categories include Agentic AI Tools and Agentic AI Applications, which require systems to observe, plan, act, and learn, thereby generating unique data challenges. ClickHouse's real-time analytics engine is designed to handle these high-volume data requirements, supporting rapid iteration and enterprise governance. Several companies, including Langfuse, LangChain’s LangSmith, Modal, Temporal, Property Finder, and Memorial Sloan Kettering, utilize ClickHouse for AI observability, evaluation, and analytics, demonstrating its capabilities in managing production volumes with high efficiency and low latency. As agentic systems demand stronger governance and clearer observability, ClickHouse continues to invest in its infrastructure to support these needs, offering a seamless transition from prototype to production on AWS.

A recently launched managed service integrates PostgreSQL and ClickHouse into a unified data stack, allowing teams to manage transactional and analytical workloads without complex pipelines. PostgreSQL acts as the system of record, while ClickHouse handles analytics at scale, facilitated by open-source components like PeerDB for data streaming and pg_clickhouse for seamless query offloading. This setup, which can be implemented using change data capture or split-writes patterns, enables efficient real-time analytics and is suitable for various applications including retail and financial systems. The managed version simplifies deployment and maintenance by offering a comprehensive solution under one ClickHouse Cloud account, enhancing scalability and reliability. By leveraging both databases, teams can efficiently manage growing data volumes and complex queries, establishing a new standard for open data stacks where PostgreSQL and ClickHouse collaboratively fulfill different roles.

A new enterprise-grade managed Postgres service integrated with ClickHouse has been announced, designed to support real-time and AI-driven applications by unifying transactional and analytical workloads without traditional complexity. This service leverages the strengths of Postgres for transactions and ClickHouse for analytics, providing a scalable, high-performance solution with reduced operational overhead. Developers benefit from a Postgres service backed by local NVMe storage, which offers significantly faster performance for disk-bound workloads, and native CDC capabilities for syncing transactional data to ClickHouse for analytics. The unified query layer, powered by the pg_clickhouse extension, allows seamless application development that spans both transactions and analytics. The collaboration with Ubicloud ensures a high standard of performance and scalability, making the service suitable for AI-driven companies that face challenges with disk IOPS and latency. The integration also includes robust features like high availability, automatic backups, and open-source components, ensuring transparency and long-term flexibility. The service is available for private preview, initially offered on AWS with plans to expand to other cloud providers.

ClickPy, a platform for tracking Python download statistics, achieved a significant milestone by surpassing 2 trillion rows in its main dataset, reflecting the extensive activity in the Python ecosystem since 2011. This achievement underscores ClickHouse's capability to manage high-volume analytical data with minimal maintenance. In response to this growth, ClickPy revamped its data ingestion pipeline, replacing a custom script with ClickPipes to enhance reliability and maintainability. This transition involved creating a separate database for testing the new pipeline without affecting ongoing operations, ultimately streamlining data ingestion and transformation processes. While implementing these changes, discrepancies in historical data were identified, necessitating careful corrections using ClickHouse's lightweight delete and update operations to ensure data accuracy without disrupting current ingestion activities. The ongoing improvements, driven by community feedback, have made ClickPy more robust and prepared for future expansion, with new features like chart exportation via Metabase enhancing its utility for users.

LaunchDarkly has transitioned to using ClickHouse Cloud to manage its data architecture, significantly improving its ability to process high-cardinality data at scale and reducing data latency to seconds. This shift has allowed LaunchDarkly to move away from expensive and maintenance-heavy streaming pipelines and batch warehouses, enabling faster feature development and unlocking new product capabilities. The adoption of ClickHouse has simplified the data infrastructure, allowing for direct querying of raw data and reducing the need for bespoke ETL jobs. This has led to a more efficient system, providing LaunchDarkly with the ability to store, query, and correlate feature flag evaluations, product analytics, and observability signals, which enhances their ability to deliver richer analytics and data-driven product features. The flexibility and performance of ClickHouse have allowed LaunchDarkly to consolidate its data systems, paving the way for new initiatives such as Release Guardian, which aims to automate reactions to software releases by connecting feature evaluations with performance metrics. The transition has also fostered a more unified approach to data across the organization, reducing architectural complexity and increasing the velocity of feature delivery.

At the end of each year, the ClickHouse engineering team introduces new features as "Christmas gifts," which include usability improvements and long-awaited updates. Some notable developments highlighted in the blog post include the ability to highlight digit groups in query prompts for better readability, and the enhancement of numeric hints for large numbers in query results. Additionally, the optimization of the distinctJSONPaths function in ClickHouse 25.12 significantly improves performance by utilizing pre-existing metadata, resulting in processing times that are over 50 times faster compared to previous versions. Similarly, DISTINCT queries on LowCardinality columns are now optimized to leverage dictionary encoding, providing faster query execution except in cases where unique values are dispersed. Furthermore, the optimization of SEMI JOIN queries through filter pushdown in ClickHouse 25.12 allows for more efficient data processing by reducing the amount of data scanned and read from disk, achieving approximately 30% faster query execution and data read reduction.

In 2025, ClickHouse underwent significant transformations, enhancing infrastructure, expanding globally, and introducing new features to improve real-time analytics. Key developments included the launch of SharedCatalog for improved cloud-scale performance, compute-compute separation with Warehouses for optimized architecture, and vertical and horizontal scaling improvements. The company expanded its global presence with new AWS, GCP, and Azure regions, while also introducing Bring Your Own Cloud (BYOC) options for AWS. Major advancements in developer experience were made, such as enhancements to the Terraform provider, the introduction of Query API Endpoints, and AI-powered features like SQL autocomplete and Ask AI. ClickStack was launched as an open-source observability stack, and security and compliance were bolstered with a new Enterprise tier and expanded compliance offerings. Data integration capabilities were improved with ClickPipes expansions, new connectors, and language clients, alongside enhanced monitoring and observability features. Notably, ClickHouse made strides in database updates, including modern search capabilities, Apache Iceberg integration, and query performance optimizations, positioning itself for continued growth and innovation in 2026.

Ramp's transition to ClickHouse Cloud has significantly enhanced its ability to provide real-time, customer-facing analytics, particularly for large enterprise clients. Initially struggling with Postgres' limitations in handling vast transaction data, Ramp adopted ClickHouse to achieve millisecond query responses, a stark improvement from the previous 40-second delays. By implementing a Postgres-to-Kafka-to-ClickHouse pipeline, Ramp's engineers, led by Director of Engineering Ryan Delgado, developed a robust OLAP platform that supports advanced features like AI-powered spend insights and proactive budget controls. This shift not only resolved existing performance bottlenecks but also positioned ClickHouse as a cornerstone of Ramp’s analytics infrastructure, enabling sophisticated forecasting and spend management tools for over 50,000 customers. Looking forward, Ramp aims to expand its OLAP capabilities to encompass a broader range of data types, thereby continuing to scale its analytics offerings.

ClickHouse enhances the efficiency of Top-N queries by utilizing advanced optimizations that significantly reduce data processing time and resource usage. These optimizations include streaming execution, which limits memory usage by only keeping current Top-N candidates; read-in-order, which avoids sorting by reading data in an already ordered state; and lazy reading, which defers the reading of non-order columns until necessary. A new technique involves using data-skipping indexes to skip granules entirely by relying on min/max metadata, reducing the rows processed before any data is read. This method, part of a broader strategy to treat Top-N queries as metadata-driven pruning problems, improves execution speed by 5× to 10× and decreases data read by orders of magnitude. It is particularly beneficial for large tables or when caches are cold, as it minimizes unnecessary reads, preserving computing and network resources. This approach allows ClickHouse to handle Top-N queries at scale effectively, maintaining high performance even with vast datasets while integrating seamlessly with existing query optimizations.

ClickHouse has acquired Langfuse, an open-source platform specializing in LLM observability, evaluations, and prompt management, to enhance AI application monitoring and optimization. This acquisition aims to combine Langfuse's developer-focused approach with ClickHouse's fast analytical capabilities, creating a comprehensive stack for building and optimizing AI applications. Langfuse stands out in the fragmented market with its strong developer community, integration with ClickHouse, and support from Fortune 500 companies. The acquisition aligns with ClickHouse's strategy of partnering with leading open-source projects to improve AI observability. It highlights the need for AI quality monitoring to address the trust gap in AI applications, emphasizing the importance of understanding AI system performance for developers, data professionals, and business users. Langfuse will continue to operate as a standalone service, with deeper integrations planned to enhance the Agentic Data Stack, ultimately benefiting AI developers and analysts by providing crucial insights into AI interactions and performance.

ClickHouse, a leader in real-time analytics and data warehousing, has announced a $400 million Series D financing led by Dragoneer Investment Group, which will support its expansion into LLM observability and the introduction of a native Postgres service. This development comes as ClickHouse experiences rapid growth, with its customer base expanding to over 3,000 and annual recurring revenue growing by more than 250% year over year. The company has acquired Langfuse, an open-source LLM observability platform, to enhance AI application monitoring, ensuring outputs are accurate and aligned with user intent. Additionally, ClickHouse has partnered with Ubicloud to offer an integrated data stack combining high-performance analytics with transactional capabilities, making it easier for developers to build AI applications. This strategic move positions ClickHouse as a unified data platform leader, capable of handling both transactional and analytical workloads, while continuing its global expansion and strengthening its ecosystem.

The January 2026 ClickHouse newsletter highlights several advancements and community contributions in the ClickHouse ecosystem. It covers the zero-copy integration of chDB with Pandas DataFrames, significantly improving query performance by enabling direct memory sharing between ClickHouse and NumPy. WKRP's migration from TimescaleDB to ClickHouse for their RuneScape tracking plugin resulted in substantial storage efficiency and operational simplification. The newsletter also emphasizes the versatility of ClickHouse through various case studies, including solving Advent of Code 2025 challenges using only ClickHouse SQL and replacing Apache Flink with ClickHouse's Kafka engine for real-time data streaming. Additionally, the newsletter features contributions from community member lgbo, who made several performance improvements, and outlines a series of upcoming events and workshops to engage with the ClickHouse community.

Picnic leverages ClickHouse Cloud to drive real-time analytics across its extensive network of automated fulfillment centers, enhancing its ability to manage millions of unique shopping journeys efficiently. Transitioning from TimescaleDB, Picnic reduced operational complexity and expanded self-service capabilities through dbt models and SQL dashboards. The real-time insights platform, powered by ClickHouse, processes operational events to provide metrics and dashboards tailored for warehouse teams, enabling them to track supply chain operations without developer intervention. This system allows Picnic to maintain its promise of timely deliveries by offering real-time data on order status, stock levels, and workforce distribution across its diverse fulfillment environments. The platform's architecture includes Java services, RabbitMQ, Apache Kafka, and ClickHouse, with Grafana used for visualization. The integration of ClickHouse has facilitated rapid growth in dbt model deployment and improved data operations scalability. Picnic's future plans involve utilizing ClickHouse for stream processing and exploring AI applications to enhance model generation and validation, ensuring continued innovation in its analytics capabilities.

The MongoDB CDC connector from ClickPipes is now in Public Beta, allowing users to replicate data from MongoDB into ClickHouse Cloud for faster analytics on document-based data. Enhanced for broader platform compatibility, the connector now supports sharded clusters and Amazon DocumentDB, offering improved reliability and secure connectivity options such as AWS PrivateLink and SSH tunneling. It leverages MongoDB's Change Streams for real-time data replication, providing up to 100x faster analytics without affecting MongoDB's operational workloads. The connector includes advanced JSON support, ensuring high-performance queries on semi-structured data while optimizing storage costs, and offers a fully managed experience with built-in metrics and monitoring. As ClickPipes moves toward General Availability, future enhancements include parallel snapshot ingestion and integration with OpenAPI and Terraform, with usage remaining free until the GA launch.

Polymarket, a prediction market platform, faced challenges in handling growing data volumes and computationally expensive analytical workloads with PostgreSQL, which led to timeouts and resource contention. To address these issues, the team, including Senior Data Engineer Max "Primo" Mershon, decided to complement PostgreSQL with ClickHouse, a faster and more scalable data warehouse solution that could manage both internal analytics and user-facing features more effectively. The transition involved integrating on-chain data from Goldsky, web analytics, and off-chain metadata into ClickHouse, allowing for efficient query execution and the creation of dynamic, granular leaderboards. This migration improved performance by reducing load on PostgreSQL and enabling Polymarket to scale their platform, supporting features like categorical leaderboards and complex internal analyses without impacting transactional workloads. The successful implementation of ClickHouse allowed Polymarket to enhance both internal and external functionalities and prepared them for future growth and increased data demands, particularly as they expanded their user base and prepared for a U.S. launch.

HighLevel, a marketing and sales platform serving over 90,000 agencies and 3 million small businesses, transitioned its data operations to ClickHouse Cloud to address performance issues as their usage grew exponentially. Previously relying on a mix of MySQL, Elasticsearch, Firestore, and a document database, HighLevel faced challenges with slow query response times, inefficient storage, and high maintenance demands. By migrating to ClickHouse Cloud, they achieved significant storage reductions and improved query performance dramatically, reducing latency from several seconds to under 200 milliseconds. This shift enabled them to efficiently handle billions of daily events and reduce operational overhead. The migration also involved optimizing their data architecture for use cases such as lead activity, workflow logs, notifications, and real-time revenue dashboards. Implementing client-side observability and adopting best practices like avoiding update-heavy patterns and optimizing schemas further enhanced their system's efficiency and reliability. This transformation allowed HighLevel to unify its data operations into a cohesive, high-performance platform, supporting its rapid growth without the previous bottlenecks.

ClickHouse version 25.12 introduces significant performance enhancements and new features, including 26 new functionalities, 31 performance optimizations, and 129 bug fixes. Notable improvements include faster execution of Top-N queries through data skipping indexes, which reduce the number of rows processed, and enhanced lazy reading via a join-style execution model, which optimizes deferred reading of non-ordering columns. The update also features a more efficient join reordering algorithm, DPsize, for INNER JOINs, which explores richer join orders to produce better execution plans. Additionally, the version marks the text index moving to beta, allowing efficient text searches with various tokenization options. Other enhancements include supporting non-constant IN lists, introducing the HMAC function for message authentication, and new dictionary capabilities for fast key-value lookups. These improvements collectively aim to boost the efficiency and scalability of ClickHouse in handling complex analytical workloads.

ClickHouse's introduction of streaming secondary indices in version 25.9 marks a significant improvement in query execution by interleaving index evaluation with data reads, as opposed to the previous method of fully scanning secondary indices before query execution. This change allows for incremental and demand-driven index evaluation, reducing latency and memory usage. Previously, secondary indices were scanned upfront to determine which granules might contain matching rows, but this process could lead to inefficiencies, particularly with highly selective queries or those with LIMIT clauses. The new approach concurrently checks index entries and reads data, halting both processes as soon as the query's requirements are met, which eliminates unnecessary work and startup delays. Demonstrations on large datasets showed that using streaming indices can significantly speed up query execution and reduce memory usage, especially for queries that can terminate early due to LIMIT conditions.

ClickHouse, a columnar database engine, initially supported only a single primary index per table, limiting query optimization based on the table's sort order. However, recent updates have introduced lightweight projections that act as secondary indexes without duplicating full data, enhancing query performance by allowing multiple projections with different sort orders. These projections store only their sorting key and a pointer back to the base table, reducing storage overhead. They enable finer filtering through granule-level pruning, significantly speeding up query response times. A benchmark test demonstrated a roughly 90% speedup when using these projections compared to a full table scan, showcasing their efficiency in processing complex queries with multiple filters. The article also explains the introduction of new settings to control the optimization process, allowing ClickHouse to leverage all applicable projections for improved query performance.

Chartmetric, a platform for music analytics, successfully transitioned from Postgres and Snowflake to ClickHouse Cloud to improve its data processing capabilities, achieving significant enhancements in query speed and storage efficiency. This migration was driven by the need to handle an ever-increasing volume of data from streaming services and social media, which had outpaced the scalability of their previous systems. By leveraging ClickHouse, Chartmetric reduced storage needs by 10 TB and accelerated query processing, allowing them to efficiently manage a 5.5 billion-row playlist cache that ingests over 15 million new records daily. The transition involved innovative approaches like using projections for efficient data retrieval and WHERE + IN filters to optimize memory usage. These changes have enabled Chartmetric to handle around 300,000 requests per hour and maintain a robust real-time analytics infrastructure. The move has not only improved operational efficiency but also reduced costs, with ClickHouse proving to be an ideal solution for their time-series data needs. Chartmetric's experience underscores the platform's scalability and flexibility, which have become integral to their data stack, complementing other tools like Postgres and Snowflake.

ClickStack, a cloud-native observability stack, leverages ClickHouse for high-performance storage and querying of telemetry data, excelling in environments with large volumes and high cardinality. However, the operational challenges of managing extensive collector fleets at such scales are addressed by integrating Bindplane, an OpenTelemetry-native telemetry pipeline. Bindplane offers centralized management, simplifying the orchestration, configuration, and upgrading of collectors, and ensuring efficient data processing and transformation before it reaches ClickStack. This integration allows for seamless telemetry routing and enrichment, enhancing ClickStack's capabilities in log analytics, monitoring, and data analysis while reducing the operational complexity typically associated with scaling observability infrastructures. As a result, organizations can maintain enterprise-grade control over rapidly expanding workloads, making the combination of ClickStack and Bindplane a robust solution for managing large-scale observability challenges.

In 2025, the observability landscape saw significant evolution, driven by increasing data volume and cardinality, the rise of tracing as a primary signal, and the consolidation of OpenTelemetry as a standard. ClickStack, introduced by ClickHouse, aimed to make high-performance observability more accessible, leveraging HyperDX UI to optimize query execution for logs, traces, and metrics. Volume alone was no longer the main challenge; instead, high cardinality emerged as a key constraint, especially with AI workloads introducing complex telemetry. Tracing became essential for understanding complex systems, often overshadowing traditional logs and metrics. OpenTelemetry became the default for new services while older systems were gradually retrofitted, despite challenges in adoption due to complex configurations and evolving standards. Open source observability solutions gained traction as teams migrated from existing platforms, valuing cost predictability and data ownership over polished user experiences. AI-driven Site Reliability Engineering (SRE) tools showed potential but remained underdeveloped, with human judgment still crucial. Data quality became a pressing issue as teams focused on the utility and cost of telemetry signals, marking a shift from mere data collection to meaningful analysis.

chDB is a Python library that integrates ClickHouse's high-performance OLAP capabilities with Pandas DataFrames, addressing the limitations of Pandas when handling large datasets. This library allows users to execute SQL queries directly on Pandas DataFrames without the need for complex setup or data serialization, significantly improving speed and efficiency. By implementing features such as automatic DataFrame discovery and optimized string encoding, chDB minimizes overhead and leverages ClickHouse's multi-threaded execution for faster query performance. The library also supports complex data structures, such as nested JSON-like objects, and offers streaming capabilities to process datasets larger than available RAM. Recent updates in chDB v4 have enhanced output performance by achieving zero-copy integration with NumPy, further reducing query execution time compared to competitors like DuckDB. This seamless integration and high-performance gain make chDB a powerful tool for data scientists who require scalable and efficient data manipulation within the familiar Pandas environment.

ClickStack, launched in May, has rapidly evolved by integrating key features such as native JSON support, dashboards import/export, and materialized views, improving performance and functionality across the platform. The product was designed to democratize observability by offering an open-source, end-to-end experience that leverages the power of ClickHouse for storing and querying high-cardinality data. Its integration with ClickHouse Cloud allows for seamless observability and analytics workflows, eliminating the need for separate infrastructure management. Throughout the year, ClickStack introduced significant enhancements like HyperDX's UI, support for custom OpenTelemetry configurations, and innovative features such as Service Maps and Event Deltas, aimed at improving user experience and scalability. As ClickStack looks towards 2026, plans include deeper cloud integration, AI-powered tools, anomaly detection, and a fully managed ClickStack experience within ClickHouse Cloud, positioning it as a leading open-source observability solution.

AI Site Reliability Engineering (SRE) tools often fail due to reliance on outdated observability platforms that lack long data retention, high-cardinality data, and fast query capabilities, which are crucial for effective incident investigation and response. Traditional AI SRE systems, primarily built on legacy observability frameworks, struggle with finding root causes due to short retention periods, dropped high-cardinality dimensions, and slow query processing, which limits their functionality to merely summarizing dashboards rather than providing actionable insights. ClickHouse, with its scalable and efficient data storage and querying capabilities, offers a robust foundation for building an effective AI SRE copilot by enabling long-term data retention, maintaining high-cardinality dimensions, and supporting fast queries. This creates an environment where AI can assist in reducing mean time to understand (MTTU) by correlating events, recognizing patterns, and providing context-rich insights to human engineers who remain responsible for decision-making. By integrating ClickHouse, organizations can enhance their incident response strategies and transition from a reactive to a proactive reliability posture, thereby not only addressing incidents faster but also reducing their frequency through upstream analysis and prevention.