August 2025 Summaries

When deciding between Tinybird and ClickHouse® Cloud for hosting ClickHouse® databases, the primary consideration often revolves around cost, driven by differing pricing models and billing mechanisms. ClickHouse® Cloud employs a consumption-based model with charges accruing based on compute, storage, data transfer, and ingestion, offering the ability to pause services to reduce costs, while Tinybird offers a plan-based model with fixed monthly costs and auto-scaling, which abstracts infrastructure management to simplify cost predictability. ClickHouse® Cloud is recommended for teams seeking granular control and cost efficiency in intermittent workloads, whereas Tinybird caters to teams prioritizing rapid deployment and stable pricing for always-on applications. While ClickHouse® Cloud may incur additional operational overhead due to the need for custom integrations, Tinybird includes built-in API functionalities, authentication, and ingestion services, reducing development efforts. Ultimately, the choice depends on workload characteristics, budget constraints, and the team's expertise in handling infrastructure versus application-level optimizations.

ClickHouse® is renowned for its speed as an analytical database, widely adopted by companies handling large data volumes, but its self-hosting requires managing complex operational challenges. Teams often encounter difficulties when scaling production deployments, such as intricate infrastructure setups and the need for careful management of upgrades, storage, and query performance. Tinybird offers a self-managed alternative that simplifies deploying ClickHouse® by providing a managed service with abstractions and tooling to ease operational complexities while allowing users to maintain infrastructure control. This includes features like automated upgrades, ingestion optimization, and API hosting, which mitigate the need for deep technical expertise in ClickHouse® operations. Tinybird's solution is particularly beneficial for open-source companies that wish to offer a hosted service without vendor lock-in, balancing high-performance analytics capabilities with reduced operational overhead.

Ensuring the security of systems using Large Language Models (LLMs) involves addressing the significant risks of data leakage and prompt injection, which cannot be fully mitigated through prompt engineering alone. It is crucial to implement security at the data layer by enforcing row-level access control (RLAC), determining what data an LLM can access by using cryptographically signed tokens that govern authentication and authorization. This approach prevents the LLM from accessing unauthorized data, ensuring that even if a prompt is injected to manipulate the system, the model cannot leak information it is not permitted to see. The Tinybird MCP Server employs this security model by using token-based authorization to limit data access, ensuring that queries are restricted to user-specific data. This method effectively safeguards LLM-based applications from unauthorized data exposure, allowing secure and dynamic interaction with data without compromising security.

The Web Analytics Starter Kit is an open-source template designed to quickly establish a customizable analytics stack, offering developers complete control over their data with real-time SQL endpoints for immediate insights. Utilizing ClickHouse®'s web analytics functions, the kit provides features like URL parameter extraction, campaign tracking, and a newly updated dashboard that supports Core Web Vitals tracking and multitenancy for managing multiple domains. A significant addition is the "Ask AI" feature, enabling users to generate analytics tables and charts using natural language without needing SQL knowledge, powered by Tinybird's infrastructure and Vercel AI SDK. The kit's core metrics, including visits, pageviews, and session durations, are enhanced with previous-period comparisons and flexible filtering, while Core Web Vitals track performance metrics like First Contentful Paint and Cumulative Layout Shift. The template is easily deployable via Tinybird, adaptable for various custom use cases, and encourages community engagement for further development and integration.

Tinybird Code is a newly developed CLI agent designed to enhance the workflow of AI-native developers working with Tinybird projects and ClickHouse® at scale. Inspired by Claude Code, it addresses the challenges of managing large-scale, real-time data, enabling users to focus on feature development rather than the intricacies of ClickHouse® operations. Built using the Pydantic AI framework, Tinybird Code supports custom model providers and multi-agent architecture, allowing for specialized sub-agents to manage tasks such as SQL queries, data quality tests, and mock data analysis. It features two usage modes—interactive and one-shot—to accommodate different development needs and integrates error recovery and modular rendering for improved user experience. Additionally, the agent is designed with secure model provider authentication, workflow-oriented tools, and safe environment handling, ensuring efficient and secure operations within Tinybird’s ecosystem.

Ghost has partnered with Tinybird to enhance its platform with real-time, multi-channel web analytics, available in the latest Ghost 6.0 release. This integration allows both Ghost(Pro) and self-hosted users to access analytics without additional costs or setup, providing detailed, first-party analytics directly within the Ghost platform. Ghost employs a hybrid architecture where MySQL manages core content and business logic, while Tinybird processes high-volume page views for real-time analytics. The system uses a client-side tracking script for data collection, processed server-side and integrated into Tinybird’s infrastructure, offering privacy-first data handling without complex streaming setups. This setup supports both cloud and local deployments, allowing flexibility for users to choose between managed hosting or self-hosting, while maintaining feature parity across both options. Ghost’s implementation demonstrates a scalable and portable analytics solution for open-source platforms, enabling real-time insights with minimal infrastructure overhead, thus supporting a broad range of users from small blogs to major publications.

The text explores the integration of AI and language models (LLMs) with analytics, focusing on using tools like Vercel AI SDK and Tinybird MCP Server to streamline data analysis without manually writing SQL queries. It details the creation of an analytics agent that simplifies querying data from a self-hosted Ghost blog, using Tinybird for real-time analytics. The post emphasizes the importance of prompt engineering in developing effective AI agents and explains the process of building a command-line interface (CLI) to interact with the analytics agent. It also discusses the setup of Tinybird for analytics, the creation of a CLI agent using Node.js, and the handling of input/output through a chat interface. The text highlights the versatility of this approach for various real-time analytics applications, beyond just blog traffic analysis, by leveraging the tools and methods discussed.