June 2026 Summaries

Coding agents like Claude Code, Cursor, and Codex CLI effectively manage the coding aspects of developing AI applications, but the more challenging tasks arise in evaluating errors, understanding why responses fail, and adapting to rapidly changing applications. Teams typically spend a significant amount of their time—60-80%—on evaluation and error analysis, often needing to redo this work with each change in the technology stack. Datadog Agent Observability captures essential telemetry data, facilitating the analysis of prompts and responses, and supports online evaluations. By utilizing Datadog's MCP Server and Pup CLI, developers can access Agent Observability data directly from their coding agents, enabling them to classify sessions, debug production failures, and evaluate application updates using real traffic data. These tools, combined with Agent Skills, allow for efficient error analysis and evaluation workflows, providing actionable insights and experiment metrics directly within the coding environment. The integration of these tools streamlines the process from investigation through to remediation, ensuring that AI applications are continuously improved in a structured and efficient manner.

Alexis Lê-Quôc highlights the limitations of off-the-shelf AI models in addressing complex, domain-specific challenges, emphasizing the importance of tuning and refining models with real-time organizational data for sustained AI value. The announcement of Adaptive ML joining Datadog marks a significant step towards this vision by integrating Adaptive ML's platform, which allows enterprises to customize and deploy specialized AI agents, with Datadog's extensive observability data expertise. Adaptive ML's Adaptive Engine enables organizations to fine-tune models using reinforcement learning and synthetic data, evaluate them with bespoke AI judges and A/B testing, and iteratively improve them by incorporating production signals. The collaboration aims to advance specialized agents and models that are more effective in production environments, leveraging real-world infrastructure, application, and security data from Datadog. This partnership seeks to explore new research opportunities in autonomous operations, transitioning from simple alert systems to adaptive, action-oriented AI systems.

Developer experience (DevEx) refers to how systems, workflows, tools, and culture influence developer productivity, and a positive DevEx can lead to organizational benefits like faster releases and reduced technical debt. However, measuring DevEx, especially in the AI era, can be challenging due to the temptation to rely on individual metrics such as lines of code or AI token consumption, which can misrepresent true productivity and widen the gap between perceived and actual contributions. Surveys indicate that a significant number of developers feel that current metrics do not accurately reflect their contributions or address their pain points. Effective measurement should focus on system health and workflow-level metrics rather than individual outputs, incorporating perceptual data to capture developer sentiment and friction points. AI adoption is often mistakenly equated with efficiency, but real productivity gains require a deeper understanding beyond mere usage statistics. Organizations should involve developers in the measurement process to foster trust and focus on metrics that genuinely impact developer satisfaction and productivity, adapting to the evolving role of AI in software development.

Memorandum M-26-14 from the Office of Management and Budget represents a significant shift in U.S. federal cybersecurity policy, moving from prescriptive logging requirements to a maturity-based framework that emphasizes continuous monitoring, threat detection, and forensic readiness. The memo establishes two key objectives: continuous event monitoring (CEM) and threat hunting, investigation, response, and forensics (THIRF), requiring agencies to centralize visibility and improve their response to security threats. The implementation timeline includes several milestones for achieving different levels of logging maturity, demanding centralized access to telemetry data and automated response capabilities. Datadog aids agencies in meeting these requirements by offering a unified observability and security platform that facilitates centralized telemetry, quick threat detection, and efficient response workflows. Datadog's tools, certified under FedRAMP High, allow agencies to enhance their cyber resilience while adhering to stringent government security standards.

Datadog for Government has achieved GovRAMP High authorization, enhancing its position as a secure observability and security platform for state, local, tribal, territorial, and education (SLTTE) organizations facing increasing complexity in technology environments. This authorization builds on Datadog's FedRAMP High certification, offering a standardized security framework aligned with NIST 800-53 Rev. 5 controls, which is crucial for agencies modernizing their mission-critical systems amidst growing cybersecurity demands and budget constraints. Datadog helps these agencies consolidate various monitoring tools into a unified platform, providing comprehensive visibility and improved operational resilience across infrastructure, applications, and cloud environments. The platform's features include centralized dashboards, network path visualization, log management, database monitoring, and AI observability, all designed to reduce complexity, enhance performance, and ensure reliable digital services. As government entities continue to modernize, Datadog's commitment to the public sector is marked by its efforts to support sensitive workloads and improve service delivery through continuous monitoring and proactive troubleshooting capabilities.

Engineering teams managing high-volume log sources like CDN edges and streaming platforms face challenges in balancing log retention costs with accessibility for investigations. Datadog's Observability Pipelines and Archive Search offer a solution by allowing teams to route raw logs to low-cost object storage while sending high-signal events to Datadog for monitoring. This approach reduces indexing costs and maintains accessibility for historical investigations without the need to switch analytics tools. Observability Pipelines come with prebuilt parsing logic, enabling the normalization and routing of logs across different CDN providers, while Archive Search facilitates querying of archived logs directly within Datadog, streamlining the investigation process. By integrating these tools, teams can maintain a seamless workflow, leveraging existing identities and access controls for both real-time monitoring and historical data analysis, ultimately enhancing their ability to act on insights efficiently.

Datadog's introduction of Datadog Kubernetes Autoscaling (DKA) has significantly optimized its Kubernetes resource management by intelligently automating both horizontal and vertical workload scaling. The implementation has helped eliminate over $3 million in annual idle compute costs, while improving reliability across its services. A central part of this success was demonstrated by Rapid, a platform team at Datadog, which adopted DKA to manage over 1,800 services and 20,000 deployments. DKA's multidimensional scaling mode addressed previous challenges with separate autoscaling tools, simplifying configurations and reducing both overprovisioning and underprovisioning issues. This allowed Rapid to fine-tune resource allocations based on actual usage, cutting costs by over 50% in some data centers and improving service reliability. DKA's successful rollout not only streamlined resource management but also fostered a culture of cost ownership among Datadog's engineering teams, enabling them to optimize services and share improvements company-wide, creating a cycle of continuous enhancement.

Security teams face the challenge of translating diverse telemetry data from sources like identity providers, cloud platforms, and web application firewalls into a unified security context for threat detection. This process is often hindered by vendor-specific languages that describe similar attack tactics, techniques, and procedures (TTPs) in different ways, leading to delays in investigations. Observability Pipelines addresses this issue by using MITRE ATT&CK Enrichment Packs, which automatically tag security events with relevant ATT&CK tactics and techniques, providing a shared framework for describing attacker operations. These packs enrich logs before they reach security tools such as SIEMs and data lakes, allowing for faster detection, threat hunting, and incident response. The enrichment process standardizes logs across various sources, ensuring that detection rules can be applied consistently. This facilitates quicker investigations, as enriched logs carry uniform fields that highlight security-relevant events, enabling teams to efficiently filter and analyze data without manual correlation. These enrichment packs are available at no additional cost and can be integrated into existing Observability Pipelines setups to enhance log context and speed up security operations.

Feature flag migrations are often seen as risky endeavors, but with careful planning and understanding of potential pitfalls, they can be managed effectively. The key challenges include ensuring logic parity between systems, maintaining configuration synchronization, and confirming cutover safety. In large organizations, legacy systems often span numerous microservices, making migrations seem daunting. However, by running parallel feature flag systems, companies can perform migrations incrementally, reducing the risk of critical incidents. Before starting, it's crucial to audit existing flags, categorize them into zombies, short-lived, and long-lived types, and establish a clear migration plan. Redirecting new flags to the new system and phasing out short-lived flags gradually decreases reliance on the legacy system. For long-lived flags, verifying logic parity through shadow mode testing or using OpenFeature's ComparisonStrategy ensures that both systems produce identical evaluations. By tracking mismatch metrics on a dashboard, teams can monitor trends and decide when to cut over safely. Freezing flag configurations shortly before cutover further ensures the integrity of the migration, making the transition smoother and less risky.

Datadog Agent Observability provides a solution for operationalizing evaluation frameworks like DeepEval and Pydantic Evals, which are often challenging for AI teams to scale beyond local experimentation. While open-source libraries offer flexibility, they require custom integration code that can be cumbersome and non-scalable. In contrast, SaaS platforms may sacrifice flexibility for convenience. Datadog addresses these issues by allowing teams to run their existing evaluations natively, maintaining the integrity of open-source frameworks while offering infrastructure for continuous evaluation runs and trace-linked regression visibility. This integration ensures that evaluations remain connected to production traces, allowing for real-time monitoring and quality assurance across development and deployment stages. By linking eval scores to production data, teams can quickly identify and address regressions as they occur, enhancing the reliability and effectiveness of their applications.

As organizations deal with scaling AI and agentic workloads, they face challenges with managing log volumes, which can impact affordability and searchability. Datadog and ClickHouse have partnered to address this issue by introducing two new capabilities, allowing teams to route high-volume logs to ClickHouse via Datadog Observability Pipelines and search these logs directly from the Datadog Log Explorer without re-ingesting them. ClickHouse, a high-performance, open-source columnar database, excels in handling large-scale, high-cardinality telemetry data with its ability to perform sub-second analysis and reduce storage costs. The integration allows teams to strategically route application and infrastructure logs for real-time monitoring or long-term compliance storage efficiently. The Federated Logs feature enables seamless querying of ClickHouse-stored logs from within the Datadog Log Explorer, providing a unified search experience. These capabilities, currently in Preview, are designed to enhance log management by optimizing cost-effectiveness and maintaining searchability across extensive datasets.

Datadog Code Threat Detection is an innovative tool designed to identify and mitigate source code attacks on GitHub pull requests (PRs) by evaluating them for suspicious behaviors such as obfuscated code, dependency abuse, credential exfiltration attempts, and CI/CD workflow tampering. Leveraging AI-assisted analysis, it examines a PR's full diff alongside repository metadata and dependency context, offering visibility into code changes that exploit trusted development workflows. This detection capability extends beyond traditional vulnerability scanning, addressing a critical part of the software supply chain that attackers increasingly target. Integrated with Datadog’s Workflow Automation, Code Threat Detection generates security signals in the Datadog Signals Explorer, allowing security and engineering teams to investigate, escalate, or remediate threats using existing tools and workflows. This approach ensures comprehensive protection from source code attacks before code reaches the production environment, enhancing the security of automated developer workflows and AI coding assistants.

Synthetic Monitoring through Datadog has introduced AI-assisted triage to streamline the process of handling synthetic test failures, helping teams distinguish between real regressions and test misconfigurations more efficiently. This feature includes AI failure summaries that classify failures as likely regressions or misconfigurations, providing insights and recommended actions based on analysis of test failures, error details, and request/response data. For instance, a Browser test failure due to outdated credentials is flagged as a misconfiguration, allowing for quick resolution without escalation. The process is further enhanced by Bits Investigation, which correlates telemetry data to generate root-cause hypotheses, enabling faster identification of issues by analyzing APM traces, logs, and infrastructure metrics. This allows teams to launch investigations directly from alerts, facilitating immediate remediation efforts. Additionally, Bits Investigation offers flexibility in configuring automatic or manual investigations based on the criticality of monitors, ensuring that response workflows align with operational priorities, and thus minimizing the time spent on manual correlation of data and enhancing incident response efficiency.

Datadog Disaster Recovery (DDR) is designed to ensure continuous observability during widespread infrastructure disruptions by allowing teams to configure a secondary Datadog site that mirrors the primary one and activates on demand. This approach addresses the challenges of cloud provider outages, as demonstrated by historical incidents involving AWS and Google Cloud, which led to significant financial losses and operational blind spots. DDR supports both active-active and active-passive configurations, with the latter being more cost-effective for most organizations. It utilizes managed resource synchronization to keep the secondary site prepared, enabling seamless failover activation without the need to rebuild resources like dashboards and monitors. Failover can be triggered through agent-based or DNS-based methods, with the flexibility for teams to decide the optimal timing based on operational needs, ensuring that telemetry data routes efficiently to the secondary site when necessary.

Datadog's recent enhancements focus on expanding comprehensive observability across all layers of a technology stack, from frontend applications to physical networks. By integrating OpenTelemetry, Datadog offers seamless instrumentation for diverse environments, including new cloud platforms like Alibaba and OVHcloud, and provides deeper insights into Kubernetes, Windows hosts, and Azure-managed services. The platform's updates include advanced features for Application Performance Monitoring (APM), infrastructure monitoring, and network device monitoring, with a strong emphasis on vendor-neutral OpenTelemetry pipelines. Datadog introduces tools like Service Remapping for consistent telemetry data management, Single Step Instrumentation for Windows hosts, and broader integrations with platforms like Vercel, Nutanix, and Oracle Fusion. These capabilities aim to enhance full-stack visibility, optimize telemetry management, and streamline root cause analysis, while new integrations extend observability to modern SaaS platforms and enterprise ecosystems.

Datadog's Synthetic Monitoring landing page, currently in Preview, enhances reliability by centralizing proactive monitoring tools into a single, actionable interface. This new feature allows engineering teams to quickly identify and address critical issues by consolidating alerts, test results, and performance data, thus reducing the time spent tracking down context across disparate dashboards. The landing page includes an Availability Overview map that visualizes real user traffic patterns to identify coverage gaps and optimize test coverage, ensuring that critical user flows are monitored. Additionally, the System Signals view aggregates performance data from frontend, backend, and connectivity layers, highlighting potential weaknesses and bottlenecks throughout the application stack. This comprehensive visualization helps teams anticipate and address issues before they impact users, such as pinpointing backend services with slow response times that could degrade the user experience. The landing page is designed to transform synthetic monitoring from isolated tests into a cohesive view of system health, enabling teams to act swiftly and prevent issues from affecting users.

Detection-based security, which relies on pre-defined rules to identify threats, can be limited in addressing the full spectrum of sophisticated AI-driven attacks. Proactive threat hunting, which involves searching for adversary behavior before alerts are triggered, is essential but challenging due to the expertise and resources required. Bits Threat Hunting, an autonomous agent within Datadog Cloud SIEM, facilitates this process by conducting hypothesis-driven investigations and analyzing telemetry data to detect anomalies and emerging threats. It enhances threat hunting coverage without increasing the workload on analysts, and integrates seamlessly with Datadog Cloud SIEM to incorporate layered threat intelligence and insights from both internal and external sources. This approach allows security teams to adapt their operations to dynamic environments and specific organizational risks, utilizing AI-driven reasoning to identify suspicious activity that may not be captured by existing detection rules. Integrations with Recorded Future and Spur Intelligence, along with the ability for organizations to bring their own threat intelligence, further enrich the platform's capabilities, making it a comprehensive tool for modern security operations.

The IDP Homepage, part of Datadog's Internal Developer Portal, serves as a central hub for engineers to prioritize and streamline their daily tasks by integrating development workflows with operational context. By consolidating information from various systems such as GitHub and Jira, along with operational data like service health and alerts, the homepage helps reduce context switching and enables developers to focus on tasks with the highest operational impact. This platform bridges the gap between developer portals and observability workflows, allowing engineers to evaluate code changes alongside service health and prioritize work accordingly. Additionally, the homepage can be customized with Datadog App Builder or Datadog Apps, enabling teams to embed custom applications tailored to their specific processes and workflows, enhancing the ability to manage deployments, approvals, and monitor alerts efficiently within a unified platform.

Security teams are facing new challenges as AI-assisted development accelerates code deployment, leading to an increased volume and speed of attacks. Datadog's 2026 State of DevSecOps report highlights that 40% of running services have exploitable vulnerabilities, with AI enabling attackers to rapidly exploit them. To address these challenges, Datadog has developed a platform providing real-time visibility into production environments, allowing teams to prioritize vulnerabilities effectively and focus on business-critical threats. Their new Runtime Prioritization Engine leverages live traces, logs, and service dependencies to assess the significance of vulnerabilities, while Bits Code and AI-native Static Code Analysis streamline development workflows by providing context-specific security insights. Furthermore, Datadog AI Guard offers runtime protection for AI agents, identifying unsafe actions and unprotected agents to prevent threats like prompt injection and data exfiltration. The Datadog platform integrates security with observability, facilitating seamless threat detection, investigation, and containment through tools like Bits Security Analyst and Bits Threat Hunting, thereby enhancing security responses and reducing manual intervention.

AI coding agents have significantly expedited the development of internal tools, particularly operational apps designed to support software release, troubleshooting, and maintenance processes. Datadog Apps offers a streamlined, code-first approach for building these applications directly within the Datadog ecosystem, allowing teams to create, deploy, and manage apps using familiar technologies like Vite and React, without needing to learn new frameworks. This integration ensures that apps benefit from existing authentication, access controls, observability, and connections, embedding seamlessly into Datadog's dashboards, notebooks, and other features. The platform also simplifies app connectivity through the Actions platform, facilitating access to external systems while maintaining consistent authentication and governance models. By inheriting Datadog's governance and observability capabilities, applications can be securely monitored and audited, with all user activity integrated into Datadog's Audit Trail. Additionally, Datadog Apps provides prebuilt components that align with its design systems, ensuring native integration and reducing context switching by embedding operational applications within the workflows they impact.

Scaleway, a European cloud and AI provider, offers an extensive range of cloud services, including high-performance GPUs, managed GenAI inference, Kubernetes, managed databases, data warehousing, and object storage. The integration with Datadog allows teams to centralize their monitoring by bringing Scaleway's telemetry data into Datadog Log Management, facilitating the collection and analysis of logs from Scaleway Cockpit and Audit Trail. This integration enables users to correlate events with infrastructure metrics and traces, providing a comprehensive view of Scaleway's infrastructure alongside other cloud providers. Teams can deploy the Datadog Agent on Scaleway instances to gather host metrics and monitor Kubernetes workloads, using prebuilt dashboards and monitor templates to quickly set up alerts and visibility. This setup aids in identifying infrastructure issues and supports compliance and security investigations, especially for organizations adhering to GDPR-aligned data residency requirements within the EU.

Datadog Journey Monitoring provides a centralized platform for digital teams to evaluate user experiences by integrating product analytics, real user monitoring (RUM), and synthetic monitoring into a single hub, allowing product, engineering, and reliability teams to assess user behavior and technical performance simultaneously. This tool helps teams quickly identify and address issues affecting critical user journeys, such as sign-ins, checkouts, and content streaming, by offering insights into conversion rates, availability service level objectives (SLOs), and technical errors. By displaying a comprehensive view of user paths and their interactions with the application, teams can pinpoint the root causes of problems without losing the broader journey context. Journey Monitoring adapts to new product flows by continuously analyzing real user activity to identify emerging critical paths, ensuring that monitoring efforts align with evolving user behaviors. Currently available in preview for existing Datadog customers, Journey Monitoring aims to provide shared business and technical insights, enhancing teams' ability to maintain and improve application health and user satisfaction.

Bits Database Optimization offers a comprehensive solution for query optimization by automating the detection, validation, and implementation of query improvements. It begins by identifying slow queries using Database Monitoring (DBM) and simulates fixes against a replica of the user's schema to ensure improvements are effective without requiring direct access to the production database. The tool leverages synthetic data to mimic real-world conditions, ensuring that recommended changes genuinely enhance performance. Once validated, Bits AI automatically creates a pull request on the exact line of code responsible for the query, embedding simulation results for easy review by developers. This streamlined process allows teams to optimize queries efficiently, reduce latency, and maintain a continuously tuned database system with minimal manual intervention.

OVHcloud, a European cloud provider popular among organizations in regulated sectors, offers a robust infrastructure that ensures data residency within Europe. However, its logs and metrics often reside outside the main observability tools used by organizations, creating challenges in cross-cloud investigations. Datadog's integration with OVHcloud resolves this by connecting OVHcloud to a unified observability platform. This integration allows centralized log collection from the OVHcloud Logs Data Platform (LDP), enabling comprehensive search, alerting, and analysis in Datadog Log Management. By installing the Datadog Agent on OVHcloud instances, organizations can gather host metrics, Application Performance Monitoring (APM) traces, and container monitoring data, which facilitates efficient root cause analysis without switching tools. Prebuilt dashboards and monitors offer immediate visibility, converting raw telemetry data into actionable insights, while three monitor templates help detect spikes in error rates and critical log counts. This integration extends Datadog's coverage to OVHcloud infrastructure, providing consistent visibility across multi-cloud environments and supporting compliance reporting for organizations relying on OVHcloud.

Agent Console, currently available in preview from Datadog, provides engineering organizations with a comprehensive view of AI coding agent activities across platforms like GitHub Copilot, Claude Code, and others. This tool allows leaders to track who is using these agents, how effectively they enhance software development, and how AI expenditure correlates with engineering output. It features real-time telemetry data, enabling leaders to monitor adoption trends, spending patterns, and the overall impact on software delivery. Additionally, Agent Console helps detect and rectify common inefficiencies such as skipped checks and retry loops by offering vetted remediation strategies. Through detailed analytics focused on adoption, velocity, and stability, the console provides insights into productive usage versus inefficient workflows, thereby assisting leaders in making informed decisions regarding AI agent integration and utilization across teams.

Infinite Cardinality Metrics is a novel framework designed to address the challenges of capturing, exploring, and scaling custom metrics in modern, highly dimensional workloads. This approach allows teams to gather comprehensive data without the need to constantly assess the cost of each new dimension, as it aligns expenses with data volume instead of cardinality. By enabling teams to track metrics like request latency across various dimensions such as service, region, and user without additional costs, Infinite Cardinality Metrics fosters a more intuitive relationship between system growth and observability costs. This system supports agentic querying and exploration, allowing engineers and AI agents to interact with complex datasets without discarding valuable context, ultimately enhancing real-time monitoring and debugging. With its emphasis on preserving valuable context and facilitating deeper visibility, Infinite Cardinality Metrics offers a transformative way to manage observability in dynamic environments, and it is now available for teams seeking to improve their infrastructure monitoring capabilities.

Datadog's latest DASH releases integrate AI advancements into engineering workflows, enhancing code writing, issue investigation, and system operation. Central to these updates is Bits Chat, a conversational AI interface available across platforms like Datadog, Slack, and mobile, allowing natural language queries for faster data analysis and action without switching tools. New features include voice input, natural language prompts for dashboard and notebook creation, and sophisticated query generation in the DDSQL Editor. Bits Chat also assists in automated trace analysis, cost anomaly investigation, and incident response, enhancing efficiency and control for engineering teams. Other enhancements include AI-assisted Grok parsing for log management, and the integration of AI tools with Datadog's observability stack, offering streamlined workflows, improved visibility, and actionable insights. These innovations provide a cohesive framework for leveraging AI in software development, enabling teams to optimize performance, manage costs, and maintain control over critical decisions.

Distributed .NET applications on Microsoft Azure often utilize managed services like Azure Service Bus, Event Hubs, Cosmos DB, and Azure API Management to facilitate the movement of requests between systems, but visibility can be challenging when issues arise. Datadog extends distributed tracing for .NET applications on these Azure services, allowing engineers to follow requests across the application flow without modifying code. This capability enables teams to trace Azure messaging services, monitor Cosmos DB operations, and connect frontend and backend services through APIM, providing a unified view to troubleshoot latency and routing issues effectively. Datadog's automatic instrumentation simplifies the setup process, reducing the need for extensive code changes and operational overhead, while enabling comprehensive visibility into the request lifecycle across Azure-managed infrastructure. This enhanced visibility helps identify bottlenecks, troubleshoot incidents from a single interface, and understand the movement of requests in complex distributed systems, making it easier for engineers to maintain application performance and reliability.

Datadog's Browser Profiler is a tool designed to help frontend engineers identify and address performance issues in web applications by tracing slowdowns back to the specific JavaScript code responsible. By integrating with Real User Monitoring (RUM), Browser Profiler captures and analyzes method stack frames from actual user sessions, aiding teams in pinpointing root causes and addressing systemic bottlenecks. It allows engineers to investigate slow interactions, compare profiling data across deployments to ensure fixes improve performance, and focus on functions that degrade user experience. The tool's Profiling Explorer aggregates data across sessions to uncover recurring patterns, and its Compare view facilitates before-and-after deployment analyses to verify improvements. Additionally, Browser Profiler's integration with Bits Chat and the Datadog MCP Server streamlines the investigative process by allowing engineers to access profiling data and insights directly through conversational interfaces. Currently in Preview, it also plans to support mobile profiling for iOS and Android in future releases, offering a comprehensive solution for enhancing frontend performance.

Bits Infrastructure Operations by Datadog is an advanced tool designed to autonomously detect, investigate, and remediate common infrastructure issues across various environments, including hosts, Kubernetes, serverless functions, and network infrastructure. It aims to alleviate the burden on infrastructure teams by automatically resolving issues like disk saturation, CrashLoopBackOff errors, and expiring TLS certificates before they escalate into incidents. The tool allows application engineers to safely address infrastructure issues affecting their services while platform engineers maintain control through defined guardrails. These guardrails set operational boundaries, ensuring safe remediation actions based on the environment and resource type, while a human-in-the-loop workflow allows teams to approve high-priority fixes. Additionally, Bits Infrastructure Operations assists teams in preventing recurring issues by learning from previously approved fixes and updating guardrails for future autonomous remediation. It also extends into the pull request workflow to flag risky infrastructure-as-code changes before they reach production, using real-time telemetry data to assess potential impacts. By reducing repetitive operational work, Bits Infrastructure Operations enables platform teams to focus on systemic improvements, ultimately enhancing the overall reliability and performance of their infrastructure.

Patterns in Datadog Agent Observability offers a sophisticated method for understanding user interactions with LLM-powered applications in production by automatically clustering these interactions into thematic groups. This approach helps identify unexpected user behaviors and interaction patterns that might not have been considered during preproduction testing. By providing a hierarchical view of production behavior, Patterns highlights operational metrics such as traffic volume, latency, cost per interaction, and error rates, allowing teams to pinpoint anomalies and potential issues. It aids in recognizing shifts in user expectations and agent behavior, thereby facilitating targeted investigations and improvements in application performance. By focusing on real production data rather than synthetic tests, Patterns ensures that evaluation coverage aligns with actual user interactions, helping to prioritize quality enhancements based on genuine usage patterns.

Datadog Network Monitoring streamlines the process of diagnosing and resolving network-related issues by integrating application, network, and configuration data into a single workflow. It enables application teams, site reliability engineers (SREs), and network specialists to quickly correlate latency alerts with specific network paths and devices, pinpoint the problematic hop, and investigate device health metrics using Network Device Monitoring (NDM). The platform also offers Network Path comparison views for historical context and Network Configuration Management (NCM) to trace issues back to recent configuration changes. Bits Investigation provides plain-language explanations of network data, making it accessible to non-network engineers, while the platform's unified interface allows for immediate remediation actions without needing to exit the system. This comprehensive approach reduces mean time to resolve (MTTR) by eliminating the need for manual data correlation and separate tool usage, thereby enhancing collaboration across teams.

Nebius AI Cloud is a comprehensive platform designed for efficiently training and deploying AI models, offering features like on-demand and reserved GPU clusters that combine high performance with cloud-native ease. The integration with Datadog enhances visibility and centralizes telemetry data from Nebius, allowing teams to monitor GPU compute, training jobs, inference services, and LLM applications across a single platform. By deploying the Datadog Agent on Nebius compute instances, users gain insights into infrastructure metrics, application performance, and GPU health, enabling proactive management of resources and troubleshooting. Datadog’s GPU Monitoring and Agent Observability further aid in identifying issues such as thermal throttling or memory errors, while also facilitating the tracing of LLM applications to optimize performance. With out-of-the-box dashboards and monitoring templates, the setup is streamlined, helping teams quickly gain actionable insights into their AI workloads without the need for extensive configuration.

The Datadog Model Context Protocol (MCP) Server introduces a robust Kubernetes toolset, providing AI agents with read-only access to Kubernetes resources and enriched context from Datadog, without direct cluster access. This toolset addresses gaps in kubectl's capabilities by enabling agents to efficiently search, inspect, and manage Kubernetes resources across multiple clusters. The tools include features for cross-cluster searches, resource context inspection with additional metadata, and token-aware manifest retrieval, allowing platform and SRE teams to streamline investigation workflows, enhance incident triage, and enforce organizational standards. By integrating these tools, teams can efficiently manage Kubernetes environments, conduct thorough analyses, and ensure compliance with governance requirements, all while minimizing manual interventions and optimizing resource utilization.

The OpenTelemetry (OTel) gateway deployment pattern is designed to help platform teams manage and scale telemetry data collection by centralizing processing tasks and routing data to observability backends. However, this approach can complicate troubleshooting due to the involvement of multiple intermediary systems, leading to increased operational overhead and slower incident resolution. Datadog's Fleet Automation offers an end-to-end view of OTel gateway architectures, aiming to streamline the troubleshooting process by unifying visibility of traffic patterns, Collector configurations, and active monitor signals. This integration enables teams to minimize context switching, quickly isolate issues, and remediate problems more efficiently by providing a cluster-level visualization of the entire OTel gateway deployment. This comprehensive view helps teams validate data routing, detect traffic anomalies, and investigate Collector issues with a cohesive workflow, ultimately enhancing the scalability and flexibility of OTel deployments while reducing operational complexity.

Bits Release is an innovative release validation and rollout intelligence system designed to handle the increasing volume and speed of code deployments, particularly those driven by AI-generated contributions. It seamlessly integrates with existing tools like Datadog and CI/CD systems to track changes from pull requests to production, ensuring that features are rolled out safely and efficiently while minimizing risks. By generating and continuously executing validation plans for each change, Bits Release detects and addresses silent production failures early, often before users experience them. It provides real-time insights into deployment performance and automatically investigates and fixes issues through Bits Dev, thereby reducing the need for manual coordination across teams. Furthermore, the system turns temporary validation checks into permanent guardrails, enhancing long-term monitoring and reliability. Ultimately, Bits Release offers engineering teams a comprehensive safety loop, allowing them to manage deployments at AI speed while maintaining high-quality standards.

Datadog Service Remapping offers users the ability to manage and standardize service names across a distributed system without requiring code changes or redeployments, enhancing the correlation of traces, logs, and metrics. This feature addresses the fragmentation caused by inconsistent service naming across different parts of the system, allowing for a unified view by using existing infrastructure tags to define service names. Service Remapping enables teams to rename services, merge or split them as needed, and safely preview the impact of changes on monitors and dashboards, maintaining seamless integration with Datadog's observability tools. This capability is particularly beneficial for teams using Single Step Instrumentation or managing diverse codebases, as it simplifies the process of achieving consistent telemetry without enforcing a full Unified Service Tagging implementation.

Datadog Forms and Case Management streamline the handling of work requests by replacing fragmented processes with an automated, centralized system. By utilizing forms with required fields, conditional logic, and automatically populated values, teams can efficiently collect and prioritize requests from both internal and external users, whether for bug reports, IT access, or vulnerability tracking. Case Management serves as a built-in ticketing system that organizes and tracks these requests, enhancing collaboration among engineering, support, and product teams while incorporating automations to reduce mean time to resolution. The platform allows for customized workflows and analytics, enabling teams to quickly respond to and manage requests, monitor trends, and ensure the right information reaches the appropriate teams. With integration capabilities and configurable workflows, Datadog enables users to automate repetitive tasks, escalate urgent issues, and maintain centralized visibility over their request processes, ultimately fostering a more cohesive and efficient operational environment.

Rufina Mariam discusses the rapid growth of telemetry volumes and how teams often employ multiple tools to manage this increase, leading to potential fragmentation and inefficiencies. The text highlights the challenges associated with using disparate systems such as Datadog for observability, Databricks for long-term storage, and ClickHouse for analytics, which can result in duplicated logs and context switching when managing logs across platforms. Federated Logs by Datadog offers a solution by allowing teams to query external data stores from a single interface without needing to switch tools, thereby maintaining control over log data across various destinations. This is exemplified in scenarios such as investigating payment failures in an e-commerce platform or analyzing user actions in a checkout process without the need for manual context switches. Additionally, Observability Pipelines play a crucial role by centralizing and standardizing log processing before data reaches its destination, supporting seamless investigations across platforms. The combined use of Federated Logs and Observability Pipelines ensures that teams can manage their telemetry data effectively while preserving their preferred storage architectures, ultimately enhancing the efficiency and comprehensiveness of log investigations.

Bits Testing, now available in Preview, offers a tool that automates the generation and maintenance of synthetic tests, alleviating the time-consuming process of manually updating test scripts for new user journeys or interface changes. It features an agent that explores applications autonomously, identifying and creating test suites for critical user journeys by providing a URL or a goal in natural language. This tool supports both deterministic tests, which follow a specific sequence for consistent application flows, and Goal-Based tests, which adapt dynamically in AI-powered and evolving environments, ensuring validation despite changes in interface or execution paths. By combining these approaches, Bits Testing maintains stable coverage for critical workflows while adapting to nondeterministic user experiences, thus reducing manual maintenance and enhancing test coverage for both stable and dynamic user journeys. To experience these benefits, users can join the Bits Testing Preview or sign up for a free trial with Datadog.

Rashel Hoover and Michael Bevilacqua-Linn discuss the complexities of AI agent development, highlighting the limitations of coding agents like Claude Code and Codex in handling tasks beyond coding, such as error analysis and dataset creation, which require human judgment. Bits Evals, a set of features in Preview, aims to automate repetitive tasks in the AI agent development loop while keeping engineers in control of critical decisions, allowing teams to resolve production failures and implement improvements more rapidly. By integrating user feedback directly into traces, Bits Evals enhances error analysis by providing a more accurate reflection of user experiences beyond operational metrics. It offers structured root cause analysis and generates evaluators based on real production behavior, streamlining workflows and reducing manual setup for evaluations. The system connects various stages of the development process into a cohesive workflow, utilizing Datadog Agent Observability to monitor and refine AI agents continuously. This approach allows teams to focus on high-priority issues, aligning their efforts with actual user outcomes and facilitating more informed deployment decisions.

Datadog's DASH keynote unveiled its innovative Bits AI support, designed to enhance both developer and operations workflows by automating detection, validation, and remediation of full-stack issues. The keynote highlighted numerous new capabilities, including AI-driven release validation, unified journey monitoring, and a new generation of Bits AI agents that preserve operational knowledge within predefined guardrails. Key features such as Bits Detection and Bits Remediation offer autonomous issue detection, investigation, and resolution, while Bits Network Device Health correlates signals across network devices for prioritized issue resolution. Datadog also introduced tools like Federated Logs and Infinite Cardinality Metrics to streamline data management and analysis. The platform further supports automated code fixes and synthetic test coverage, enabling teams to manage AI applications effectively and maintain end-to-end observability, scalability, and security. By offering solutions like Datadog AI Guard and the Runtime Prioritization Engine, Datadog ensures robust security and efficient vulnerability management, allowing teams to deploy AI agents swiftly without compromising safety.

Bits Data Analysis is a new feature in preview from Datadog that allows teams to obtain fast and reliable answers to business questions by leveraging curated context from their data infrastructure and the wider Datadog ecosystem. Integrated with various data tools like Snowflake, BigQuery, and Tableau, it uses governed metadata and lineage information to automatically build context, thereby eliminating the need for manual semantic-layer work. By correlating business metrics with application telemetry data, Bits Data Analysis can trace production root causes, such as identifying latency spikes in checkout services that correlate with revenue dips. It operates seamlessly across platforms like Slack, the Datadog web app, and coding agents, ensuring consistent answers across all queries. The Context Workbench enables data teams to oversee and refine the context used for answers, ensuring accuracy and consistency. This feature empowers different organizational functions to self-serve analytics, enhancing their ability to make data-driven decisions.

Datadog is modernizing its API authentication model to better accommodate the diverse and complex access patterns of contemporary infrastructures, which now include CI/CD pipelines, Terraform workflows, and AI agents operating across teams. The new model introduces four purpose-specific authentication approaches: Personal Access Tokens (PATs) for developer workflows, Service Access Tokens (SATs) for automation and non-human workloads, workload identity federation for cloud-native workloads, and OAuth clients for delegated and temporary AI agent access. This revamped system replaces the dual-key authentication model, providing scoped, identity-aware authentication that enhances security, compliance, and audit attribution by offering tailored credential options based on the specific user or system requirements. While the existing application keys will remain operational until after Q3 2026, Datadog encourages users to transition to the new authentication methods to leverage improved features such as better scoping, clearer identity, and proper time-to-live support.

Datadog has introduced a range of new features and integrations to enhance security and compliance across modern systems, facilitating AI-assisted investigation and remediation throughout the security lifecycle, including code, cloud, APIs, and sensitive data. Key advancements include AI agents for automated threat hunts, expanded SIEM capabilities, and a modernized approach to authentication and governance. Datadog now enables dynamic routing for security notifications, automated bidirectional ticket creation, and deeper security integrations with over 30 new additions since the last DASH event. The platform also supports automated investigations via standalone solutions like Bits Security Analyst, while introducing tools for faster remediation of code and cloud risks with AI-native SAST and Code Threat Detection. Furthermore, Datadog enhances compliance monitoring and security posture with integrations in Oracle Cloud Infrastructure, FedRAMP High certification for government workloads, and secretless authentication for Azure environments. These innovations aim to provide security teams with streamlined workflows, higher accuracy in threat detection, and improved visibility across multi-cloud environments.

Datadog End User Device Monitoring (EUDM) introduces the Command Center, a centralized tool designed to help IT teams manage endpoint issues across large fleets of employee devices. By leveraging Bits Investigation, the Command Center identifies, prioritizes, and presents fleet-wide endpoint problems through concise, human-readable cards that include severity indicators, root cause descriptions, and recommended actions. This system automates the correlation of device metrics, application logs, and network data, thereby reducing the time and expertise required to address issues. It also integrates with external systems like Jira or ServiceNow to track remediation efforts and provides a comprehensive view of common scenarios affecting network performance, device health, and AI tool usage. The Command Center aims to streamline the entire process from detection to resolution, offering a repeatable workflow that enhances response efficiency and reduces manual investigation.

Bits Code is an innovative tool designed to streamline the remediation process for engineering teams by integrating with Datadog to convert high-impact errors and other findings directly into reviewable pull requests. This tool addresses the traditional time-consuming workflow of error detection, investigation, and code modification by leveraging Datadog's telemetry data to investigate issues, generate targeted code changes, add unit tests, and facilitate the continuous integration process before opening a pull request in GitHub. Bits Code operates within enterprise access controls, supports automation for repetitive tasks like fixing flaky tests and clearing dependency advisories, and provides flexibility with scheduled or on-demand runs. It emphasizes grounding code changes in the production context by using real-time telemetry, which reduces the risk of inaccurate fixes and prioritizes issues based on their impact. Engineers maintain control over code review decisions, ensuring that human oversight is preserved while increasing the efficiency of the remediation workflow. Available to all Datadog customers, Bits Code also supports GitHub Cloud with beta support for GitLab, allowing teams to enhance their error resolution processes while maintaining robust governance and privacy controls.

Bits Detection is a feature within Datadog that automatically creates, tunes, and maintains monitoring coverage for services by leveraging existing telemetry, ownership, and deployment data, as well as the source code when connected. It addresses the issue of coverage drift by continuously updating detection logic based on observed production behavior, rather than relying on static thresholds, allowing teams to identify critical service paths and reduce mean time to detection. By focusing on endpoint-level monitoring and adapting to changes in service infrastructure, Bits Detection helps organizations manage the complexities of modern systems and ensures that monitoring remains aligned with production. This system not only identifies issues earlier but also facilitates the investigation and remediation process by connecting detected issues to relevant telemetry and offering autonomous triage and recovery actions, thus reducing the manual workload associated with traditional monitoring practices.

AI agents are becoming increasingly integrated into the software development life cycle, aiding in tasks such as code generation, debugging, and incident management, but they often require context-switching due to a lack of direct access to observability data. To address this, Datadog has introduced integrations with various AI chat and coding agents, including Claude Code, ChatGPT, and Codex, through its MCP Server. These integrations provide a secure and structured interface for AI agents to query Datadog’s observability data and tools using natural language, facilitating tasks like log searches, incident lookups, and service governance without leaving their current environment. By supporting a range of platforms and offering seamless connectivity, Datadog enhances collaboration and efficiency for engineers, allowing them to manage resources, investigate production issues, and analyze telemetry data directly through their preferred AI interfaces. The integrations are designed to work across different environments, whether local or remote, employing OAuth for authentication and ensuring consistency in accessing core capabilities, while also supporting multi-org setups and role-based access control.

Engineers are leveraging AI tools like Claude, Cursor, and ChatGPT, integrated with Datadog's MCP Server and Apps, to streamline incident investigations and data analysis by providing interactive visualizations directly within AI conversations. This setup allows users to view real-time Datadog graphs, monitors, and analytics widgets without leaving the AI interface, enhancing workflows by enabling the exploration of telemetry data, user behavior, and latency issues in a more intuitive and seamless manner. By embedding these interactive elements in AI chats, users can conduct in-depth analyses, ask follow-up questions, and correlate quantitative trends with user behavior, all within a single conversational context, thereby improving efficiency and maintaining investigation continuity. Datadog MCP Apps facilitate this integration, offering features like timeseries visualizations, session replays, and monitor cards, which help developers and product teams quickly diagnose and resolve issues from within their existing AI-powered workflows.

Datadog's latest DASH announcements introduce a suite of tools designed to enhance cost management, governance, and reliability for large-scale operations, focusing on reducing cloud and AI expenditures, enhancing observability, and streamlining management. Key features include Datadog Disaster Recovery (DDR) for maintaining observability during cloud outages, Remote SDK Upgrades for easier updates across services, and Organization Groups to manage multiple Datadog organizations as a single unit. Cloud Cost Management centralizes AI spending from multiple providers, enabling better tracking and optimization, while Cost Optimization Automation executes recommendations to reduce infrastructure spending. Additionally, Datadog's updates facilitate improved incident management and workflow automation through features like the IDP Homepage, Datadog Forms, Case Management, and enhanced integrations with Microsoft Teams and Slack, supporting seamless communication and coordination during incidents. Cloudcraft's live diagrams for Oracle and VMWare infrastructures offer real-time insights into system architecture, aiding teams in incident response and cost optimization. These advancements aim to simplify complexity and control costs while maintaining operational efficiency and reliability at scale.

Datadog's Data Observability and Jobs Monitoring features offer a comprehensive solution for optimizing large-scale data processing jobs running on Apache Spark and Databricks, which often face challenges related to extended durations and high costs. By leveraging Datadog's tools, data engineering teams can proactively identify optimization opportunities, generate actionable recommendations, and apply fixes directly within the Datadog UI, all while integrating seamlessly with existing AI-assisted development workflows. This process involves the use of Datadog MCP Server, Bits Code, and Spark execution context, enabling engineers to focus on specific job performance issues without being overwhelmed by extensive Spark history data. The platform provides prioritized recommendations across teams and workloads, such as adjusting Spark configurations, refining code and queries, or selecting appropriate infrastructure, with each suggestion tied to potential cost savings. By combining proactive recommendations with reactive investigation tools, Datadog enhances the efficiency and efficacy of data pipeline optimizations, ultimately achieving significant reductions in compute costs and job durations for engineering teams.

Datadog's new feature, Network Path, combined with End User Device Monitoring, allows IT and network teams to trace network paths from an end user's device to a SaaS application, offering detailed visibility into each hop with metrics on latency and packet loss. This comprehensive tracing tool helps pinpoint where slowdowns occur, enabling more precise root cause analysis by distinguishing whether issues arise from local networks, ISPs, or SaaS provider infrastructure. Additionally, it facilitates comparison of network paths across devices and over time, providing insights into whether issues are persistent, intermittent, or isolated to specific contexts, thereby reducing the need for manual log correlation. By enabling direct network path analysis from a user's device, Datadog centralizes telemetry and accelerates the investigation of performance issues, offering a streamlined approach to troubleshooting by connecting device-level and network-level signals.

The Claude Compliance API, integrated with Datadog Cloud SIEM, offers enterprises a comprehensive solution for monitoring and managing security and compliance activities related to Claude Enterprise. This integration allows organizations to ingest, normalize, and analyze compliance events, such as user access, API key management, and configuration changes, in a structured manner that aligns with Datadog's security data model. By utilizing prebuilt detection rules, teams can identify potentially risky activities, like unexpected administrative changes or suspicious access patterns, and correlate them with broader security telemetry to enhance context and reduce false positives. The integration provides dashboards for real-time monitoring and exploratory analysis, enabling security teams to quickly identify anomalies and investigate compliance events alongside other security data. This unified approach not only streamlines security operations but also supports governance initiatives by offering detailed insights into Claude Enterprise activity across an organization.

Bits Chat is a conversational interface integrated into Datadog's web and mobile applications and collaboration tools like Slack, designed to enhance efficiency by allowing users to search, visualize, and act upon data using natural language without switching between tools. It assists engineers in navigating Datadog's environment by quickly surfacing relevant dashboards, monitors, logs, and telemetry data in response to natural language queries, thus streamlining the troubleshooting process. Additionally, Bits Chat facilitates onboarding by providing contextual best practices and guidance directly within the workflow, and it can generate and modify dashboards and notebooks based on user prompts, promoting seamless collaboration. By correlating signals from various sources, it accelerates issue resolution and helps document findings, thereby minimizing the time required to connect data insights and take action. Bits Chat is now generally available, offering a more integrated approach to managing and analyzing data within Datadog.

Bits Agent Builder is a newly introduced tool by Datadog that enables users to create custom AI agents capable of analyzing data, making complex decisions, and adapting to changes in real time without requiring extensive hardcoding. These agents can interpret Datadog's observability data and third-party signals, allowing them to take automated or on-demand actions based on specific goals and data sources. The tool supports chaining multiple agents for complex workflows, enabling them to specialize in distinct steps while maintaining overall process automation and control. Agents can leverage over 2,000 prebuilt actions from Datadog’s Action Catalog and connect to external services like Kubernetes and cloud platforms. Examples of its application include automating investigations of application errors, responding to critical vulnerabilities, and generating service documentation. By integrating Bits Agent Builder, teams can streamline operations, reduce manual work, and enhance their decision-making capabilities using AI-driven automation.

In AWS environments, data perimeters serve as a protective measure to ensure that cloud identities access only trusted resources from authorized networks, with controls applied at various infrastructure levels. Despite the potential of organization-level solutions like resource control policies (RCPs) and service control policies (SCPs), many organizations still rely on per-resource controls, which can create security gaps. The text discusses how these gaps can be tested using Stratus Red Team, an open-source threat emulation tool, and highlights scenarios such as visibility, identity, network, and resource perimeters. It emphasizes the importance of organization-level controls to close these gaps, illustrating how per-resource policies can be vulnerable to attacks such as policy drift or unauthorized access. Additionally, the text provides examples of how to configure and use Stratus Red Team to simulate attacks, underscoring the need for robust logging and monitoring to detect unauthorized activities and address potential security risks effectively.