April 2026 Summaries

Grafana Assistant is an agentic observability tool that enhances troubleshooting efficiency by preemptively learning about a user's infrastructure, thus eliminating the need for engineers to repeatedly provide context during incident responses. The assistant autonomously builds and maintains a knowledge base of the infrastructure by continuously analyzing data from Prometheus, Loki, and Tempo sources, creating a comprehensive understanding of service identities, metrics, deployment topologies, dependencies, and log structures. This preloaded knowledge allows for rapid and precise responses to queries, significantly reducing response times and increasing accuracy, especially in complex environments where team members may not be fully familiar with all systems. The process is automated, requiring no configuration or maintenance from the user, and respects access controls to ensure data security. This functionality is seamlessly integrated into Grafana Cloud, offering users a straightforward way to leverage their existing telemetry data for improved incident management.

The gcx CLI tool is designed to enhance observability for engineers by integrating Grafana Cloud functionalities directly into the command line, thereby improving the efficiency and accuracy of code management and system monitoring. As coding increasingly relies on command-line interface (CLI) tools, gcx addresses the visibility gap that exists when agents, although efficient at generating code, lack insights into the production environment. By enabling agents to access real-time data and production metrics, gcx allows for quicker incident resolution and informed decision-making. It offers features such as instrumentation, alerting, SLO management, and frontend and backend observability, all of which can be managed as code. The tool is particularly optimized for agent-driven environments and supports seamless integration with command-line tools like git and kubectl, providing a stable and efficient way for agents to operate without the need for additional layers or integration tools. This setup empowers agents to perform tasks traditionally reserved for human engineers, thus bridging the gap between development and production environments and reducing the time and effort required to maintain system health and performance.

Secrets management for Grafana Cloud k6 has been introduced to secure performance testing by allowing users to store and use sensitive data such as API keys and credentials safely, without hardcoding them into scripts. This feature helps manage the sprawl of sensitive information as testing suites expand, reducing the risk of exposure and simplifying test maintenance. Users can create, edit, and delete secrets through the Grafana Cloud UI, where secret values are kept write-only to prevent accidental exposure. In tests, secrets can be accessed through the k6/secrets module, enabling seamless integration into existing scripts and ensuring sensitive data is protected even during execution. This functionality is available in public preview for Grafana Cloud k6 and is also generally available in Grafana Cloud Synthetic Monitoring, offering a secure and streamlined approach to performance testing.

Cloud Provider Observability in Grafana Cloud offers users the ability to customize preconfigured dashboards and views for AWS, Azure, and Google Cloud, enhancing the utility of these dashboards by allowing personalization to fit specific workflows and preferences. Users can connect existing dashboards, generate new ones with AI, and edit instance drill-down views to tailor their observability experience within the app. The customization process is centralized on the configure page of each cloud service, where users can set default dashboards, create quick links, and modify panels and queries, ensuring a consistent user experience across various observability surfaces within Grafana Cloud. This flexibility allows users to maintain the prebuilt views while integrating their own or AI-generated dashboards, providing a comprehensive and adaptable monitoring solution.

o11y-bench is an open-source benchmark designed to evaluate AI agents in observability workflows within a real Grafana environment, specifically focusing on tasks like querying metrics, logs, and traces, investigating incidents, and making dashboard changes. Built on the Harbor framework, it provides a standardized environment for testing, helping users discern between seemingly effective agents in demos and those truly reliable in real-world scenarios. The benchmark includes 63 tasks across different observability domains, such as Prometheus, Loki, Tempo, and dashboard management, and uses metrics like Pass^3 and Pass@3 to measure consistency and success rates. By open sourcing the tasks, environment, and grading logic, o11y-bench aims to be transparent and reproducible, encouraging community engagement to advance agent capabilities in observability. The initial benchmark trials showed that while many models could succeed at least once in three attempts, only a few demonstrated consistent reliability, emphasizing the importance of measuring reliability over occasional success in observability tasks.

Grafana Labs has acquired Logline to enhance its log management system, Loki, by improving the execution of complex "needle-in-the-haystack" queries and full-text search capabilities. Logline, founded by Jason Nochlin, offers a novel indexing approach designed for high-cardinality attributes over object storage, which significantly reduces the amount of data scanned in searches, thereby speeding up query performance without compromising Loki's cost-effectiveness and simplicity. This strategic acquisition aligns with Grafana Labs' mission to make observability more accessible and cost-efficient while maintaining a strong commitment to the open-source community. Initial benchmarks demonstrate a dramatic reduction in data scanned for specific queries, exemplified by a reduction from 3.5 TB to 8 GB in a UUID search. The new capabilities are currently available in limited private preview on Grafana Cloud Logs, with plans to extend them to Loki OSS users in the next major release.

Pyroscope 2.0 marks a significant advancement in continuous profiling, offering a rearchitected solution that enhances speed, cost-effectiveness, and operational simplicity while supporting the OpenTelemetry Protocol for profiling. With continuous profiling becoming integral to the observability stack, Pyroscope 2.0 addresses the inefficiencies of its predecessor by eliminating write-path replication and reducing the symbolic storage footprint, which makes profiling at scale more feasible by significantly lowering storage and compute costs. The new architecture facilitates faster root cause analysis and improved query performance by making the read path stateless, allowing for scalable processing based on demand, thus optimizing resource use during bursty access patterns. These changes were pressure-tested and refined in Grafana Cloud, where Pyroscope 2.0 has been in production, processing vast amounts of profiling data. The cleaner architecture not only resolves previous operational challenges but also enables new features like metrics from profiles and richer query types, making continuous profiling more accessible and practical for teams.

Grafana is evolving its Assistant, a purpose-built LLM agent, to be more accessible and customizable for diverse environments, as announced during GrafanaCON 2026. Previously available only in Grafana Cloud, Assistant is now extended to Grafana Enterprise and Grafana OSS users, allowing them to connect from self-managed environments and use it for real-time data analysis, dashboard building, and incident resolution. This expansion aims to incorporate AI into software development workflows by minimizing context switching and offering new integrations with tools like Slack, Microsoft Teams, and APIs, thereby enhancing observability practices. Users can customize Assistant through "skills," which guide agents with specialized knowledge and automate tasks, while new features like Assistant automations and a remote hosted MCP server broaden its functional scope. The initiative underscores Grafana's commitment to providing flexible AI-driven solutions that cater to varying organizational needs and promote efficient software development cycles.

GrafanaCON 2026 in Barcelona unveiled significant updates for Grafana, including the latest Grafana 13 release, which enhances data insights through suggested and dynamic dashboards, as well as the introduction of the AI-powered Grafana Assistant for customized dashboard creation and streamlined SQL expressions. The event highlighted improvements in operating Grafana at scale, such as Git Sync for GitOps workflows and the Grafana Advisor tool for system performance. AI Observability in Grafana Cloud, now in public preview, offers real-time monitoring of AI agent behaviors, while the open-source o11y-bench helps evaluate AI agents on observability workflows. Additionally, updates to open-source projects like Loki and Pyroscope 2.0 focus on faster, more efficient logging and profiling, respectively, while k6 2.0 introduces AI-driven performance testing. The event also launched the Grafana Marketplace for plugin distribution and celebrated the Golden Grot Awards, recognizing innovative dashboards and contributions to AI observability.

AI Observability in Grafana Cloud is a newly launched solution designed to address the challenges of monitoring AI systems, which traditional observability tools struggle to handle. As organizations transition from cloud-native to AI-native environments, understanding AI agent behavior becomes critical, as these agents make decisions and interact in complex ways. Grafana Cloud's AI Observability allows users to observe AI agent behavior in real time, evaluate outputs, detect anomalies, and correlate agent sessions with traditional telemetry signals, all within the same environment. Built to extend the capabilities of Grafana's existing observability platform, it helps teams understand AI performance, spot issues, and address potential data exposures by integrating seamlessly with OpenTelemetry. The tool also provides detailed insights into agent interactions, allowing for efficient debugging and performance optimization. AI Observability includes features like user annotations and alerting, and is available for use in Grafana Cloud, offering a comprehensive solution for monitoring AI workloads at scale.

Grafana 13, unveiled at GrafanaCON 2026, offers a streamlined and flexible platform to enhance data visualization and operational efficiency. Key features include pre-built dashboards tailored to specific data sources, visualization suggestions leveraging metadata, and a revamped query editor for easier complex panel creation. The release also introduces Grafana Assistant, an AI-powered tool, and new visualization options like the Graphviz panel plugin. Dynamic dashboards provide a more intuitive and scalable navigation experience, while Git Sync and Grafana Advisor improve operational management and instance health. Additionally, the introduction of the Grafana Marketplace allows for the distribution and sale of plugins, and a new Enterprise data source integration with IBM DB2 expands data visualization capabilities. These updates aim to empower users to gain actionable insights from data quickly and manage Grafana at scale effectively.

Grafana Labs has introduced the Grafana Marketplace, a platform that allows independent software vendors, systems integrators, and developers to sell and distribute plugins for Grafana, as announced during GrafanaCON 2026. Designed to support the sustainable growth of the Grafana plugin ecosystem, the marketplace offers a path for third-party developers to earn revenue while ensuring ongoing updates and support for their plugins. While emphasizing that community and open-source plugins remain central to Grafana, the marketplace provides an opt-in model for paid plugins, offering users diverse options to meet their observability needs. The initial phase of the marketplace features contributions from founding partners like Crest Data, Phenisys, and KensoBI, who bring specialized solutions to the Grafana ecosystem. Developers interested in participating in the Grafana Marketplace can join the pilot program, which offers additional support and reduced fees for early participants, aiming to foster a vibrant community for creating and sharing observability tools.

Grafana Cloud offers a Databricks integration designed to provide comprehensive visibility into Databricks workloads, catering to the distinct needs of FinOps, platform, SRE, and analytics teams. This integration streamlines monitoring by importing metrics from Databricks workspaces directly into Grafana Cloud, eliminating the need for custom exporters or building dashboards from scratch. It includes three prebuilt dashboards that offer insights into costs, job reliability, and SQL warehouse performance, along with 14 alerting rules aimed at notifying relevant teams about significant changes or issues. The integration utilizes an open-source exporter, databricks-prometheus-exporter, which accesses Databricks System Tables for billing, job, and query performance data. Set up requires a Grafana Cloud account, Databricks credentials, and configuration through a setup wizard, with considerations for billing data lag and specific table permissions. The service aims to simplify the monitoring process and is part of Grafana Cloud's broader suite of tools for managing metrics, logs, and traces, with a free tier available for new users.

GrafanaCON 2026 is set to take place in Barcelona from April 20-22, celebrating the Grafana community and its open-source ecosystem with technical deep dives, hands-on demos, and hackathon projects. The event, located at Palau de Congressos de Catalunya, promises sessions led by industry leaders like Google and LEGO, focusing on AI tools and open-source observability. Highlights include Grafana Labs' big announcements, over 30 sessions covering updates like Grafana 13 and AI advancements, and community-led talks. Attendees can participate in hands-on labs, the Science Fair, and the Golden Grot Awards, while engaging in informal discussions at Birds of a Feather sessions. The event also offers group discounts for registration and ensures an inclusive atmosphere with a Code of Conduct for all participants.

Grafana Alerting introduces alert enrichment, a feature designed to enhance the context of alerts in Grafana Cloud by attaching relevant information, such as log lines, annotations, and links to dashboards, to facilitate faster and more effective incident response. This enrichment aims to transform alerts from mere notifications into comprehensive entry points for investigation and resolution, allowing engineers to prioritize incident management over data gathering. The update addresses the limitations of previous Grafana alerts, which could only present information derived directly from query expressions, often resulting in complex and costly queries. Enrichment supports various functionalities, including automatic investigations through Grafana services like Assistant and Sift, and the integration of external data sources, improving the operational effectiveness of alerts. By treating enrichment as a standard practice rather than an optional feature, Grafana Alerting aims to provide users with precise, actionable insights, enabling them to quickly identify and address issues without unnecessary context switching. The feature can be configured at the rule level or globally, providing flexibility in how enrichments are applied across different alerts.

The Kubernetes Monitoring Helm chart version 4.0 marks a significant update designed to address user challenges in monitoring setups, offering enhanced predictability, flexibility, and maintainability. The update transitions from list-based to map-based configurations for destinations, allowing more precise overrides without reordering issues. The restructuring also introduces customizable collector mapping, separating telemetry services deployment, and distinct configurations for cluster, host, and cost metrics, reducing unnecessary deployments and memory usage. The profiling feature is now modular, enabling users to activate specific profilers as needed. These changes aim to simplify configuration management, optimize resource allocation, and provide greater transparency in the setup of monitoring tools within Kubernetes clusters. A migration tool is available to assist users in transitioning from previous versions, reflecting these structural improvements.

Profiles Drilldown's integration with Grafana Cloud Knowledge Graph offers a streamlined, queryless approach to diagnosing performance bottlenecks by directly linking profiling data, such as CPU and memory usage, to existing observability workflows. This integration allows users to seamlessly explore code-level details through an intuitive interface, without the need to switch tools or craft complex queries, thus accelerating root cause analysis. By connecting metrics, logs, traces, and profiles across various observability solutions in Grafana Cloud, the knowledge graph creates a comprehensive map of a system's components, enabling faster and more efficient issue resolution. The integration ensures contextual insights by automatically filtering profiling data based on the entity being investigated, enhancing the troubleshooting process by allowing users to view flame graphs and isolate costly operations directly within the knowledge graph. This advancement reflects Grafana's commitment to reducing complexity in observability workflows, making it easier for users to identify and understand performance issues in their systems.

Managing synthetic monitoring checks as code using Terraform with Grafana Cloud offers a scalable and consistent approach for teams dealing with numerous checks across multiple environments. This method allows for programmatic definition, version control, and consistent deployment of checks, overcoming challenges like inconsistent configurations and lack of change tracking inherent in manual UI-based management. Grafana Cloud Synthetic Monitoring provides a blackbox monitoring solution, facilitating various checks to measure service availability, latency, and correctness. The process involves prototyping checks in the Grafana UI, configuring the Terraform provider, exporting checks to Terraform, and then managing and updating them through Terraform, which becomes the source of truth. By adopting this approach, teams can ensure identical check configurations across environments, facilitate collaboration through pull requests, and easily scale operations. The use of Terraform not only supports efficient monitoring management but also integrates seamlessly into CI/CD pipelines, providing a robust framework for maintaining and auditing monitoring setups as systems grow.

Grafana Cloud's Private Data Source Connect (PDC) offers a secure solution for modern businesses to integrate proprietary datasets from private networks into their observability platforms without compromising security. By deploying a lightweight PDC agent that creates an encrypted SSH tunnel, businesses can securely access and visualize relational data such as business metrics directly in their Grafana dashboards. Grafana Assistant, an AI-powered tool, enhances this capability by translating natural language prompts into complex SQL queries, enabling users to transform PostgreSQL data into meaningful visualizations without needing SQL expertise. This integration not only simplifies the process of querying and visualizing data but also expands the use of observability tools beyond engineering, allowing businesses to track metrics related to security, compliance, and customer behavior. The setup process is automated through a Terraform blueprint, which facilitates secure infrastructure provisioning and includes a demo using the World Happiness Report dataset to showcase the potential of combining PDC with PostgreSQL and Grafana Assistant for advanced business analytics.

Grafana Cloud's "query fair usage" policy allows users to query up to 100 times their monthly ingested log volume in gigabytes without additional charges, designed to prevent misuse from resource-heavy queries while maintaining cost predictability and infrastructure protection. Users can monitor their query usage through the Grafana dashboard, with options like the Billing/Usage dashboard and a newly redesigned Cost Management experience. Key practices to optimize query usage include configuring alert rules properly, using label selectors and log pipeline filters early, narrowing time ranges, and utilizing aggregation and recording rules for efficiency. Grafana Cloud also offers Loki query limit policies, currently in public preview, to provide control over query result sizes and prevent costly queries, with support and potential changes anticipated before general release. Understanding and managing query usage can prevent unexpected billing surprises, as usage is calculated based on the larger value between ingested and queried gigabytes, impacting the monthly invoice.

In a recent episode of "Grafana's Big Tent" podcast, Mat Ryer and his team discuss the complexities of observability in Go programming, emphasizing the importance of starting with logs to derive metrics. Through conversations with experts like Donia Chaiehloudj and Charles Korn, the episode explores how logs can be transformed into metrics, the role of tracing in understanding complex systems, and the utility of pprof for performance profiling. The discussion highlights common challenges and solutions, such as using eBPF for deeper system visibility and the nuances of handling errors in Go. The podcast episode aims to provide practical insights for developers navigating the intricacies of observability while also touching on areas where Go could improve to better support debugging and error tracking.