February 2026 Summaries

In the first episode of This Month in Datadog in 2026, Jeremy discusses the protection of agentic AI applications using AI Guard, enhanced collaboration during incidents through five new Incident Management releases, and the confident deployment of software with Feature Flags. Kevin highlights the benefits of Datadog Data Observability, which helps detect data quality and pipeline issues early, providing insights into data lineage and the impact on AI models or BI dashboards. The episode also introduces the Datadog Ambassador program, upcoming Datadog Summits in London and Sydney, and various new features such as cross-product retention filters, Sensitive Data Scanner, Code Coverage, and Fleet Automation for automatic agent upgrades. This monthly roundup aims to keep viewers informed about the latest features and product announcements, encouraging them to subscribe to the YouTube channel or explore the Datadog platform through a free trial.

Datadog's release of Toto, a state-of-the-art time series foundation model, emphasizes its strong zero-shot forecasting capabilities, which have been well-received with over nine million downloads. However, in real-world applications, the availability of historical data and additional predictive features can significantly enhance forecast accuracy. To address this, Datadog has introduced fine-tuning capabilities and support for exogenous covariates, allowing users to adapt Toto to specific workloads and incorporate known future predictors, such as billing cycles or marketing campaigns, to improve forecast precision. These enhancements, demonstrated through cloud cost forecasting and public benchmark datasets, show that fine-tuning with exogenous covariates notably improves performance across various metrics. The new features are available in the Toto repository on GitHub, and Datadog encourages users to experiment with different settings to optimize forecasting accuracy for domain-specific needs.

Achieving end-to-end observability in applications is a significant challenge due to the need for coordinated frontend and backend monitoring, which often requires collaboration between different teams. Datadog addresses this issue by enabling seamless integration of Real User Monitoring (RUM) and Application Performance Monitoring (APM) for Java applications with a single command, eliminating the need for frontend code changes. This integration automatically instruments Java applications using Datadog's single-step instrumentation (SSI) mechanism, which loads the necessary SDKs and RUM JavaScript snippet into the application processes. By capturing both frontend user session data and backend service traces, Datadog correlates these insights to provide a comprehensive view of application performance, enabling faster troubleshooting and root cause identification. This unified approach enhances the efficiency of DevOps and SRE teams by reducing the complexity and friction typically associated with achieving full observability, ultimately improving the ability to address issues affecting user experiences.

Amazon Machine Images (AMIs) are crucial templates for launching and scaling EC2 instances, and their management significantly affects the cloud attack surface. AMIs can be sourced privately as golden images, from public repositories, or from community sources, each presenting different security risks and trust boundaries. The article discusses how improper management of AMIs can lead to vulnerabilities such as insecure configurations, data leaks, and exposure to supply chain attacks. For instance, using unvetted public AMIs can introduce risks like embedded crypto miners or name confusion attacks, where malicious images mimic legitimate ones. To mitigate these risks, best practices include standardizing secure defaults using CIS benchmarks, routinely scanning AMIs for vulnerabilities, maintaining an inventory of active instances, and enforcing controls such as restrict searches by publisher and creating allowlists of approved images. The text also highlights the role of Datadog's security tools, like the whoAMI-scanner, in identifying and addressing unverified community AMIs, thereby helping organizations to prioritize remediation and reduce their cloud environment's exposure to potential threats.

Datadog Real User Monitoring (RUM) offers a comprehensive solution for monitoring mobile app launch performance, specifically targeting challenges that arise from varying startup experiences on iOS and Android platforms. By capturing detailed metrics like Time to Initial Display (TTID) and Time to Full Display (TTFD), RUM provides insights into how apps perform across different launch scenarios, such as cold, warm, and prewarmed launches. This nuanced approach allows app teams to make informed comparisons and optimizations, focusing on user retention and conversion. The tool's ability to classify startup metrics by launch type and provide real user data facilitates a deep investigation into performance issues, supporting organizations in monitoring and alerting on startup performance at scale. Datadog encourages users to upgrade their SDKs to version 3.5.0 or later to leverage these capabilities, offering a structured method to enhance user experience by understanding and addressing the specifics of app launch behavior.

Datadog's engineering teams have developed AI Guard, an application designed to monitor and secure AI agents by detecting and blocking unsafe behaviors in real time. The system uses LLM Observability to evaluate, iterate on, and monitor AI Guard, capturing every model decision and measuring the impact of changes with statistically valid datasets. AI Guard protects Datadog's Bits AI Agents by analyzing requests from user prompts to model responses, identifying potentially harmful actions such as prompt injections or unsafe code execution. The development process includes creating a comprehensive dataset that combines real and simulated agent activity, enabling accurate detection models trained on edge cases. AI Guard's performance is continuously evaluated using structured experiments, with every change to the system tracked and reviewed to ensure alignment with desired operational metrics. It currently operates on both OpenAI models and custom self-hosted models, and Datadog is exploring additional models and frameworks to expand its capabilities. In production, AI Guard integrates with Datadog's AI gateway to provide seamless evaluation and monitoring, improving the speed and efficiency of detecting and resolving issues, thus reinforcing AI system reliability.

As organizations adopt AI-assisted coding and face rapid code changes, ensuring test coverage becomes increasingly challenging, with untested code more likely to slip into repositories. Datadog Code Coverage addresses this issue by offering multi-level visibility into test coverage, allowing teams to identify untested additions from repository levels down to individual lines of code. The tool provides a unified view of test coverage across platforms like GitHub and GitLab, helping teams quickly spot consistently untested areas and revealing coverage gaps at both file and service levels. By integrating with source code repositories, Datadog Code Coverage allows developers to see line-level annotations of untested sections, enabling them to address these gaps efficiently. This level of visibility is crucial as AI-generated code paths, while syntactically correct, may lack supporting tests, and Datadog helps teams catch coverage gaps early, reducing the risk of reliability issues as the codebase evolves.

Modern CI/CD pipelines have streamlined the process of building, testing, and deploying code, yet many teams still face challenges during the manual and risky rollout phase of a release. This often involves painstakingly monitoring metrics and manually adjusting traffic, which can lead to stress and potential errors. The article discusses the benefits of automating this process through the use of guardrail metrics, which act as automated safety nets by monitoring key health indicators during a release and triggering automatic rollbacks if thresholds are breached. By integrating these guardrails with observability platforms like Datadog, teams can link releases directly to existing metrics, such as error rates and latency, ensuring a more reliable and less labor-intensive rollout process. This approach allows for progressive rollouts that automatically expand or contract based on real-time telemetry data, reducing the risk of incidents and the need for manual oversight. The article encourages teams to evaluate their release processes and consider adopting automated feature management to enhance efficiency and confidence in their software deployment.

Modern enterprises increasingly depend on software-defined wide area networks (SD-WANs) to connect their branch offices, data centers, and cloud environments, which enhances flexibility and performance but also adds complexity. To address issues such as slow applications or intermittent connections, Datadog's integration with Versa SD-WAN offers comprehensive visibility into Versa-managed environments by collecting critical metrics related to device health, service level agreements (SLA), interfaces, and traffic. This integration allows network and infrastructure teams to validate traffic paths, identify performance degradation, and understand the impact of application and user traffic across sites. The integration helps teams monitor WAN edge and controller health, track SLA metrics for links and tunnels, analyze interface utilization and congestion, and understand application, user, and Quality of Service (QoS) traffic patterns. With these insights, teams can ensure SLA compliance, optimize capacity planning, and quickly troubleshoot connectivity issues, thereby enhancing the overall performance and reliability of their networks.

Over the years, the Datadog Agent's artifact size increased significantly, creating challenges for size-sensitive contexts, but a concerted effort to optimize its Go binaries has resulted in a remarkable size reduction without sacrificing features. By systematically auditing dependencies, refactoring code, and leveraging linker optimizations, Datadog managed to reduce the size of its Go binaries by up to 77%, bringing them back to their sizes from five years ago. These efforts involved understanding the Go compiler's handling of dependencies, utilizing tools like goda and go-size-analyzer to visualize and manage binary sizes, and eliminating unnecessary dependencies. Additionally, method dead code elimination and strategic use of build tags played crucial roles in achieving these size reductions. The optimization efforts not only improved performance and reduced resource usage for Datadog's users but also contributed enhancements back to the Go community, benefiting projects like Kubernetes. This optimization journey showcases the importance of continuous improvement and collaboration in software engineering.

Datadog APM Recommendations is a tool designed to enhance application performance and reliability by analyzing telemetry data from various Datadog services, including Application Performance Monitoring, Continuous Profiler, Real User Monitoring, and Database Monitoring. It provides actionable insights by identifying performance, reliability, and resource efficiency issues, explaining their significance, and offering guidance on resolving them before they escalate. This system detects common anti-patterns, such as N+1 database queries and repeated sequential API calls, and presents them in a centralized view to facilitate assessment across teams, services, and environments. Recommendations are prioritized based on forecasted business impact and user effect, with detailed context provided to aid in triage and assigning ownership. The tool also tracks ongoing recommendations, allowing teams to monitor progress and systemic inefficiencies over time, ultimately enabling proactive optimization and reducing the risk of incidents by addressing issues before they affect users.

Security teams face the complex task of identifying and remediating vulnerabilities in applications due to the intricate web of dependencies and transitive dependencies within frameworks, SDKs, and utilities. Datadog Software Composition Analysis (SCA), part of Datadog Code Security, assists teams in addressing these challenges by providing visibility into the dependency chain, which helps identify the root library introducing vulnerable dependencies and recommends safe upgrades. The process involves detecting vulnerabilities such as Log4Shell, connecting them to their root libraries, and offering guidance for upgrades that ensure minimal disruption to the build. This comprehensive approach not only highlights the necessary changes but also considers the wider impact on the codebase, facilitating a more efficient and secure remediation process. Datadog's internal research indicates that around 70% of vulnerabilities stem from transitive dependencies, emphasizing the importance of tools like Datadog SCA in closing the gap between vulnerability detection and effective resolution.

Datadog Sheets is a tool designed to streamline the audit preparation process for security teams by providing a centralized, spreadsheet-style interface that simplifies the collection, analysis, and reporting of security vulnerabilities across various environments. It integrates with Datadog's Code Security and Cloud Security capabilities, allowing teams to gather comprehensive vulnerability data from code, infrastructure, and cloud resources in a unified view. By enabling teams to filter vulnerabilities by severity, time range, and asset type, and by offering prebuilt templates for common security and compliance tasks, Datadog Sheets reduces manual rework and ensures consistent reporting. It also facilitates correlation of findings across different sources, improving the presentation of risk data to auditors by allowing vulnerabilities to be reported by resource rather than by tool. The tool ultimately reduces audit preparation time and enhances visibility and reporting consistency for governance, risk, and compliance purposes.

As enterprises expand, identifying user-reported issues becomes more challenging, especially with the implementation of Software-defined wide area networks (SD-WANs), which, while simplifying the addition of branch offices, complicate the differentiation between connectivity degradation and application behavior changes. FortiManager offers a centralized control plane for Fortinet Secure SD-WAN, reducing operational complexity and enabling efficient monitoring of link quality and FortiGate edge device health through its integration with Datadog. This integration aids engineering teams by providing metrics, logs, and Network Device Monitoring (NDM) metadata to identify connectivity issues and application performance impacts. Datadog's dashboard facilitates visualization of link performance, helping teams diagnose network issues by correlating data like latency and packet loss, and build monitors for conditions such as jitter spikes. Additionally, visibility into FortiGate devices' resource utilization helps prevent outages by identifying resource pressures, while application and traffic insights allow for informed troubleshooting of network incidents. By centralizing link-quality metrics, device health, and application usage, the FortiManager integration enhances the efficiency of incident investigation and decision-making regarding routing policy and capacity planning.

Datadog Code Coverage is a new capability designed to help engineering and development teams maintain testing standards amid the rapid pace of AI-assisted development and frequent code changes. It provides a comprehensive view of test coverage across repositories, breaking down test activity by line, file, service, and branch to identify untested code early and enforce consistent testing standards. By integrating coverage data within Datadog's software delivery workflows, it allows for seamless access to coverage insights alongside existing metrics, traces, and CI signals. Automated test generation through Bits AI Dev Agent further strengthens test suites by filling coverage gaps without manual effort, enabling teams to focus on higher-impact work. As part of the Datadog Software Delivery Suite, this tool unifies software quality signals, supporting organizations in managing code quality even as they scale microservices or standardize testing expectations. Datadog Code Coverage is now available, offering a 14-day free trial for new customers.

Datadog Code Coverage is a tool that helps modern engineering teams maintain consistent testing standards across various services and repositories as they scale, by providing automated test coverage insights and enforcement within CI pipelines and pull requests. It addresses the challenge of inconsistent coverage standards and prevents untested code from being merged by implementing automated gates that block changes not meeting predefined thresholds. This ensures every codebase operates under a unified definition of quality, regardless of tech stack or team structure. By surfacing coverage insights directly in pull requests, developers receive immediate feedback, facilitating the addition of necessary tests before code merges. This approach allows teams to maintain development velocity without sacrificing quality, ultimately enabling faster and more reliable software delivery without the risks associated with untested code.

Datadog Observability Pipelines offers a solution for enhancing log data by integrating ServiceNow CMDB (Configuration Management Database) context, thereby streamlining incident management and security investigations. Typically, DevOps and security teams rely on the CMDB for infrastructure metadata, but logs often arrive without this context, necessitating manual correlation during incidents. Observability Pipelines solve this by using Datadog Reference Tables to automatically enrich logs with up-to-date CMDB data, such as ownership, application, and dependency information. This enrichment occurs within the user's infrastructure, allowing for flexible log processing and routing based on CMDB-derived attributes, thus eliminating the need for duplicated configuration in downstream tools. The integration helps prioritize and route events to destinations like SIEMs or data lakes based on business impact and ownership while maintaining synchronization with ServiceNow to reflect any organizational changes. Additionally, the ability to rehydrate historical logs with current CMDB data supports effective investigation of incidents and security threats.

High-performance computing (HPC) clusters depend heavily on fast and reliable shared storage to prevent CPU cores and GPUs from idling due to data delays. The integration of Datadog with Lustre, a parallel file system, offers detailed visibility into storage performance, essential for overcoming bottlenecks that slow down workloads. This integration monitors file system operations, metadata activity, I/O throughput, and overall file system health, enabling HPC teams to identify and resolve issues such as metadata bottlenecks and job slowdowns. By tracking metrics from metadata servers, object storage servers, and clients, teams can pinpoint performance bottlenecks, understand job-specific metadata pressure, and optimize storage configurations. Datadog provides a comprehensive view of cluster performance, correlating Lustre telemetry with job scheduler data and other HPC metrics, to effectively troubleshoot and enhance storage efficiency and network performance. This capability allows for informed decisions on tuning storage parameters and rebalancing workloads, thus maintaining the productivity of HPC clusters.

Datadog's Workload Protection Findings, now available in Preview, offers a dedicated space for runtime posture monitoring, separating it from threat detection to help security teams manage urgency across different concerns. Findings identifies and tracks risky behaviors that may not be immediately malicious, such as running compilers in production containers, by providing a separate view that allows teams to address these issues proactively. The feature integrates with Case Management, enabling teams to create cases, assign them, and document remediation steps directly within Datadog, facilitating collaboration among security, SRE, and platform teams. By using runtime telemetry data, Findings connects behaviors to specific hosts, containers, pods, and services, offering comprehensive visibility and contributing to a complete posture life cycle from detection to remediation and verification. This approach helps teams maintain focus during incident response while still addressing underlying risks, ensuring a robust security posture without creating visibility gaps.

Datadog Product Analytics has been updated to assist product managers (PMs) in making quick and informed decisions by simplifying user data analysis without the need for SQL or engineering input. The platform offers features like Action Management, which allows PMs to label and manage user actions with a no-code interface, enhancing data consistency and trust. This tool also enables PMs to analyze user journeys, save and share actionable insights, and quickly address common product questions using chart templates. These enhancements allow PMs to efficiently track user behavior, improve analysis speed, and maintain a shared understanding across teams, ultimately leading to more confident product decisions.

Datadog APM enhances the visibility of event-driven systems using Google Cloud Pub/Sub by offering producer-aware distributed tracing that connects traces from producers to consumers, thus resolving challenges in monitoring such systems. Through features like span links for batch publishing, detecting Pub/Sub push deliveries in Cloud Run, and preserving trace context across asynchronous acknowledgments, Datadog helps maintain a comprehensive trace of message life cycles, even when messages undergo batching or delayed acknowledgments. This setup aids in debugging production issues like partial batch fan-out, queue backlogs, and delayed acknowledgments, offering insights into end-to-end processing times and allowing more accurate performance assessments. To achieve this, users must configure their systems to ensure trace continuity, particularly by enabling metadata writing to HTTP headers for push subscriptions, which is crucial for maintaining the connection between producer and consumer traces.

In the ongoing debate over whether to use JavaScript or CSS in frontend development, the choice often depends on the specific requirements of the project. While JavaScript is favored for its robust libraries that facilitate advanced UI features, it also presents challenges in terms of performance, accessibility, and code complexity. Conversely, CSS, though less feature-rich, aligns better with browsers' native rendering processes, thereby enhancing performance and simplifying code. The discussion highlights scenarios where CSS can outperform JavaScript, particularly in text styling, transitions, animations, and layout, due to its lower processing needs and simpler syntax. The article also delves into how web content rendering impacts performance, explaining that the critical rendering path of browsers is more efficiently navigated using CSS, which is less parser-blocking than JavaScript. It underscores the use of CSS for optimizing page loading speeds, improving SEO, and offering a better user experience, suggesting that developers can leverage CSS's built-in capabilities for tasks like text styling and animations to improve overall app performance. Additionally, tools like Datadog Real User Monitoring (RUM) are recommended for identifying performance bottlenecks and areas where CSS can replace JavaScript to enhance user experience.

Datadog AI Guard is a new security feature designed to enhance the protection of agentic AI applications that utilize large language models (LLMs) by providing real-time evaluation of prompts, responses, and tool calls. It aims to mitigate risks such as misuse, data exposure, and manipulation by introducing intelligent guardrails and context-aware detection mechanisms. AI Guard offers two primary protection capabilities: Prompt Protection, which assesses prompts and responses based on context and policy; and Tool Protection, which scrutinizes tool calls to ensure alignment with intended goals. These features help prevent multistep attacks and data exfiltration attempts, allowing benign actions to proceed while blocking harmful ones. The system is integrated with Datadog's existing infrastructure, requiring no additional components, and provides a comprehensive interface for monitoring AI activity and understanding security posture. AI Guard's flexible enforcement modes enable organizations to observe, tune policies, and transition to blocking unsafe actions, facilitating a secure yet innovative environment for AI development.

Modern applications often log more information than necessary, including personally identifiable information (PII) such as human names, which can lead to privacy and compliance issues. Datadog's Sensitive Data Scanner (SDS) offers a solution by introducing human name detection using a machine learning (ML) pipeline to identify and optionally obfuscate human names in logs. This approach reduces the need for fragile regex patterns and helps organizations comply with privacy regulations like GDPR. The system allows for actions such as redaction, hashing, or masking of detected names, ensuring that sensitive information is managed appropriately without disrupting non-sensitive data. Additionally, it provides a centralized findings page for reviewing detections and supports tuning and suppressing rules to avoid false positives. By incorporating human name detection into their processes, organizations can enhance their PII compliance, apply consistent governance, and streamline the management of sensitive data in their logs.