July 2025 Summaries

Tinybird Code, a newly launched CLI agent with expertise in ClickHouse®, is designed to streamline the development, deployment, management, and optimization of Tinybird projects. The text explores its integration with Claude Code's sub-agents feature, which allows developers to assign specific workflows to specialized AI assistants. One notable use case demonstrated is the creation of an 8-bit Atari-style browser game with real-time analytics. This integration utilizes Tinybird Code as a sub-agent within Claude Code to handle analytics tasks, such as creating data sources and API endpoints, while Claude Code manages the game development. The process includes configuring Tinybird Code as a sub-agent, building and testing the game locally, and deploying to Tinybird Cloud. The multi-agent workflow exemplifies the benefits of leveraging specialized agents for domain-specific tasks, maintaining context, optimizing performance, and refining instructions for improved future workflows. Despite some challenges encountered, particularly with cloud deployment, the experience highlights the potential of AI-assisted development in creating efficient, end-to-end applications.

Creating CI/CD workflows is essential for automating the software development lifecycle, ensuring faster and more reliable releases, reducing manual errors, and enhancing team collaboration. The integration of an electricity data visualization project into Tinybird Forward demonstrates a structured CI/CD setup that includes local testing, staging deployment, and production deployment. Tinybird Forward, an evolution of the Tinybird platform, facilitates real-time data app development with features like local-first development, data-as-code, and simplified CI/CD processes. The CI pipeline manages pull requests, runs tests, and ensures code is ready for cloud deployment, while the CD component automates the promotion of changes through staging to production, ensuring only tested changes are deployed live. The workflow also includes a staging dashboard setup to visually test deployments before they go live. Tinybird Forward simplifies schema changes and enforces a data-as-code approach, offering developers a robust platform designed to prevent accidental changes and automate safe deployments, ultimately enhancing the development experience.

The article provides a comprehensive guide on setting up a local development environment for real-time analytics using Kafka and Tinybird, emphasizing the importance of a local-first development approach to enhance fast iteration and testing. It demonstrates how to configure Docker Compose with Redpanda (a Kafka-compatible broker) and Tinybird Local to create an environment that mimics production settings without the delays of deployment, allowing developers to test changes instantly and debug data pipelines in real-time. The guide also covers writing and running tests to validate both SQL logic and the Kafka-Tinybird connection, showcasing a workflow that includes a custom CI setup to ensure production-grade quality assurance. Through this setup, developers can maintain a rapid development pace while ensuring robust analytics features, making the process as seamless and efficient as regular application development.

In an effort to streamline and prioritize customer support issues, a system was developed using Plain webhooks, Tinybird, and an AI agent to automate the process of generating a weekly digest for the product team. This system captures events from Plain and pushes them into a Tinybird datasource, where the data is processed and transformed into a coherent format using materialized views. Conversations are aggregated and analyzed, with the results made accessible through a secured HTTP endpoint. A cron job further enhances the data by generating GPT summaries, which are then compiled into a Slack channel report for easy viewing. The integration of these technologies enables efficient data ingestion, real-time processing, and secure delivery of insights, allowing teams to focus on the most pressing customer issues.

On April 28th, 2025, Spain experienced a nationwide blackout lasting over 8 to 10 hours, highlighting the country's dependency on electricity and sparking discussions on the management of the electrical grid. In response, a real-time analytics project was developed to monitor Spain's electricity system, utilizing Tinybird as the analytics engine, Python for data tracking, and Grafana for visualization. The project involves collecting data from the ESIOS API, ingesting it into Tinybird, and materializing it for efficient querying to display on Grafana dashboards. Insights from the project reveal the diverse nature of Spain's electricity production, with solar energy playing a significant role during sunny days, and the country's status as a net exporter since 2021. While the exact cause of the blackout remains uncertain, it is suspected to be a combination of events, including a sudden drop in solar production and disrupted international exchanges. The project serves as a model for understanding and leveraging real-time data to enhance the monitoring and management of critical infrastructure.

Analytics agents, powered by AI, are versatile tools capable of autonomously interacting with data to generate insights, build reports, and power applications, facilitated by Tinybird's Model Context Protocol (MCP). These agents, functioning as Large Language Models (LLMs), are designed to autonomously ask and answer analytical questions by running on a loop, using tools via MCP, and can be deployed in various forms such as CLI tools, chatbots, or background processes. Building an analytics agent requires a continuous loop, an LLM client with MCP tools, and explicit task instructions, with frameworks available for multiple programming languages to aid in this process. To connect these agents to Tinybird data, users need to utilize the Tinybird MCP server and a resource-scoped token for access control, with the goal of delivering specific instructions for data collection, analysis, and reporting. The process includes collecting data from Tinybird endpoints, interpreting results for trends and anomalies, and reporting findings via channels like Slack. Examples of analytics agents include those for organization metrics, CPU spike analysis, and web analytics, each designed to offer actionable insights and recommendations based on data analysis.

The guide provides a comprehensive tutorial on building a production-ready analytics agent using the Agno framework and Tinybird's real-time analytics platform. Agno is a Python framework designed for creating AI agents that quickly instantiate and efficiently use memory while supporting multiple modalities and reasoning models. Tinybird offers the data infrastructure necessary for these agents, featuring sub-second query responses, SQL compatibility, and API generation. The guide covers step-by-step instructions for setting up a Tinybird workspace, creating data sources and API endpoints, and integrating them with an AI agent developed using Agno. It also details how to connect this agent to external services, manage its memory and context, and configure output options through channels like CLI, email, and Slack. Advanced features include persistent memory, error handling, and deployment strategies for production environments. The guide emphasizes the synergy between Agno and Tinybird, highlighting how their combined capabilities enable intelligent, scalable, and proactive data analysis solutions.