February 2026 Summaries

Engineering metrics can significantly enhance business outcomes when used correctly, but they often fail due to poor implementation and leadership practices. Two common pitfalls are the assumption that simply acquiring a metrics tool will drive change and the belief that metrics alone can fix engineering issues. For metrics to be effective, they must be executive-owned and integrated into decision-making processes, with leaders actively using and demonstrating the value of these tools. Furthermore, metrics should be seen as instruments for identifying areas of improvement rather than solutions themselves. Organizations need to establish clear definitions and a consistent operating rhythm to turn metrics into actionable insights, ensuring that any changes made are measured for their impact on outcomes. Ultimately, the goal is not to measure individuals but to assess and enhance the system as a whole, fostering a culture where metrics are used to confirm assumptions and guide strategic decisions.

Cloud cost optimization strategies often fail because they are reactive rather than proactive, focusing on reporting and cost allocation after expenses have occurred rather than integrating cost control into the engineering workflow. Traditional methods of observing and optimizing cloud costs, borrowed from on-premises capacity planning, are insufficient for the rapid provisioning capabilities of cloud environments. Effective cost management requires shifting cost awareness left, embedding it into CI/CD pipelines and infrastructure-as-code workflows, so that developers can see and address potential cost impacts before deployment. Harness Cloud Cost Management exemplifies this proactive approach by integrating cost visibility and governance directly into engineering processes, providing real-time dashboards, anomaly detection, and in-workflow cost policies that prevent wasteful configurations from reaching production. This approach emphasizes cost efficiency over mere cost reduction, allowing teams to maintain engineering velocity while staying within defined cost limits, ultimately enabling growth rather than constraining it.

Harness has transformed its Chaos Engineering module into Resilience Testing to better address modern reliability challenges in software delivery. Resilience Testing integrates Chaos Engineering, Load Testing, and Disaster Recovery Testing into a unified, continuous approach, leveraging automation and AI-driven insights to enhance system resilience. This evolution stems from the need to validate systems beyond isolated failures, focusing instead on understanding and improving system behavior under various stressors and real-world conditions. The transition is supported by Harness's acquisition of Chaos Native, which contributed the widely adopted open-source project LitmusChaos to the CNCF. Through active community collaboration and enterprise integration, Harness aims to embed resilience testing into the software delivery lifecycle, reflecting lessons learned from operating and scaling real systems.

Harness is making significant strides in enhancing its DevSecOps capabilities by introducing Artifact Registry GA for secure artifact management and integrating Qwiet AI, now known as Harness SAST and SCA, to improve security measures within its platform. The company has launched FinOps automation through Harness AutoStopping™ to reduce cloud costs and has been recognized with three major DevOps Dozen Awards, highlighting its prowess as a leading AI-native platform. Harness is expanding its AI engineering organization in India and aims to bolster its workforce to 1,000 employees, focusing on AI innovation. Additionally, Harness has raised $240 million to accelerate its AI-driven Software Delivery platform, introduced a new AI Knowledge Agent to enhance developer portals, and added Flyway support to its Database DevOps for increased flexibility. The company's efforts and customer impact have earned it a place on Inc.'s Power Partner list for 2025, emphasizing its commitment to innovation and customer success.

February's updates from Harness focus on enhancing AI security and delivery in software development, emphasizing enterprise-grade application security, API security, SRE automation, and a significant DevOps Agent upgrade. The newly available WAAP Public MCP Server facilitates API security data querying with natural language in AI environments, integrating insights into developer workflows. Harness's Security Testing Orchestration now includes AI-powered static and software composition analysis tools, prioritizing reachable vulnerabilities to reduce noise in high-speed AI coding environments. Enhancements to Jira integration in AI SRE streamline incident management by dynamically loading required fields based on project and issue type. A major upgrade to the DevOps Agent, powered by the Opus 4.5 foundation model, improves speed, context retention, and accuracy for complex pipeline automation. These innovations aim to resolve the AI Velocity Paradox, ensuring faster, safer software delivery by addressing downstream bottlenecks in testing, security, and deployment, thereby turning AI potential into tangible business velocity.

Harness has announced the General Availability of Artifact Registry, a key advancement in artifact management designed to streamline and secure software delivery workflows by integrating artifact management directly into the CI/CD platform. This innovation addresses the fragmentation and security challenges traditionally associated with managing artifacts in separate systems by embedding capabilities such as security scanning, RBAC, audit logging, and policy enforcement within the same platform where builds and deployments occur. Harness Artifact Registry supports a diverse range of container formats and package ecosystems, providing a centralized control point that enhances visibility, governance, and risk management at the earliest stages of the delivery lifecycle. The introduction of Dependency Firewall further exemplifies its commitment to security by blocking risky dependencies at ingestion, reducing the potential for supply chain vulnerabilities. This integration ultimately aims to simplify artifact management, reduce operational costs, and enhance the security posture of organizations by offering a unified platform for managing diverse artifacts.

In the transition from traditional DevOps to AI-driven delivery, IBM's Chief Architect, Devan Shah, highlights the intricate balance between speed and security, emphasizing the evolution from "DevOps to DevSecOps to AI." With IBM's "OnePipeline" platform, built on Tekton and Argo CD, the company addresses the complexities of managing over 450 developers, underscoring the importance of automated security measures and the AI Velocity Paradox, where rapid code generation can be hindered by manual security processes. AI tools, like IBM's internal agent "Bob," are integrated into the software development lifecycle with contextual rules and AI code reviews to prevent technical debt and ensure maintainability. The concept of "Crown Jewels In, Crown Jewels Out" is introduced to stress the risks of feeding sensitive data into AI models, advocating for robust Data Security Posture Management and architectural principles like Just-In-Time token provisioning to mitigate risks. Devan Shah also introduces the "No Jail" architectural principle, which focuses on a minimal set of non-negotiable security measures to balance speed and compliance, aiming to make security a foundational feature of the software delivery process rather than a hindrance.

Harness has introduced a new feature called Project Movement, allowing users to transfer entire projects between different organizations with just a few clicks, addressing a major customer need for flexibility in response to organizational changes. This feature enables pipelines, configurations, and historical data to move seamlessly, maintaining integrity and continuity without the need to rebuild from scratch, thus preventing operational downtime or data loss. While project-specific resources transfer with ease, organization-level resources such as shared connectors or secrets remain in place, requiring users to update references post-move, ensuring that security boundaries are maintained. The feature is designed to simplify and expedite the reorganization process, reducing the time and effort traditionally required for such transitions, and focuses on preserving the robustness and security of the system by maintaining strong isolation measures.

The text provides a comprehensive overview of advancements in DevOps, focusing on the transition from automation to autonomy through the use of Agentic AI, which enhances infrastructure management. It highlights innovations like CI/CD-driven NoSQL change control for improved compliance, human-aware Site Reliability Engineering (SRE), and AI-driven application security, which streamline complex post-code challenges. The document discusses challenges in scaling GitOps, specifically the "Argo ceiling," and offers solutions via a GitOps control plane. It also introduces the Harness Human-Aware Change Agent AI that connects human insight to impactful changes, alongside enriching OpenTelemetry spans with feature flag context for better performance understanding. The text further explores chaos engineering for production resilience, standardizing feature flag management through Harness FME and OpenFeature, and dynamic pipelines for adaptable CI/CD processes. Additionally, it includes predictions for 2026 on AI and DevSecOps, strategies for safe cloud migration with Harness automation, and the support of Google AlloyDB by Harness Database DevOps for secure and automated PostgreSQL delivery.

AI-assisted coding has shifted from experimental to mainstream, promising faster development and improved code quality; however, the speed gains in code creation are often negated by underdeveloped downstream processes like testing, deployment, and security, a phenomenon termed the AI Velocity Paradox. A survey of 900 engineers and technical managers across the US, UK, France, and Germany reveals that while 63% of organizations report faster code delivery due to AI, only 43% have automated their continuous integration and build pipelines, leading to increased security risks, manual efforts, and potential cost overruns. Organizations are categorized into four quadrants based on their AI and automation maturity, with "Velocity Leaders" achieving balance by integrating AI across the entire software delivery lifecycle. The report suggests that to fully realize AI's potential, companies must focus on automating downstream processes, consolidating tools, and implementing AI-driven governance, transforming the speed achieved in coding into reliable and resilient software delivery.

Harness has introduced a native MongoDB executor for Liquibase Community Edition, addressing long-standing challenges faced by users in integrating MongoDB with database workflows in a CI/CD environment without relying on paid extensions. This initiative aims to make MongoDB DevOps processes more accessible and consistent by enabling users to run MongoDB scripts, generate changelogs, and align migration workflows with CI/CD practices within a fully open-source ecosystem. The native executor enhances MongoDB adoption by removing historical limitations and fostering open collaboration, thus facilitating seamless integration with Liquibase. Additionally, Harness has provided a Python utility to generate Liquibase-compatible changelogs from existing MongoDB environments, further easing the transition to versioned workflows. This effort by Harness highlights its commitment to contributing to the open-source community, ensuring that foundational database capabilities remain accessible and driving innovation through collaborative development.

Harness Artifact Registry has evolved from merely managing artifacts to becoming a crucial control point for software supply chain security and governance. The platform has been rapidly adopted by enterprises due to its ability to prevent risky dependencies from entering pipelines, thereby reducing exposure to supply chain threats. By integrating artifact management into the Harness platform, it offers centralized visibility and control across a wide range of artifact formats, promoting a seamless developer experience without slowing down workflows. The registry's built-in Dependency Firewall proactively blocks risky dependencies before they're downloaded, and its quarantine feature ensures that only compliant artifacts are deployed. Artifact Registry is designed to integrate smoothly with existing CI/CD and security tools, enhancing early risk detection and governance. As the platform continues to expand, it aims to support more package ecosystems and further enhance its governance and security capabilities, ultimately transforming artifact management and software delivery for modern enterprises.

Continuous Integration (CI) metrics are crucial for platform engineering leaders to enhance developer productivity, control costs, and ensure reliable software delivery. By tracking metrics such as build duration, queue time, success rates, and cost per build, organizations can transform raw CI data into actionable insights that drive business outcomes. Effective metric tracking helps identify productivity bottlenecks, optimize resource usage, and maintain security standards. Harness CI demonstrates this by delivering faster builds and enabling budget predictability through intelligent caching and optimized testing. Additionally, metrics like mean time to recovery and flaky test rates help teams address build failures and improve developer confidence. Standardizing pipeline metadata and automating metric collection are key steps towards operationalizing these insights, ensuring teams focus on metrics that reveal bottlenecks, control costs, and maintain pipeline reliability.

Unit testing is a crucial aspect of continuous integration and continuous deployment (CI/CD) pipelines, offering fast, isolated validation of individual components in software development. By employing strategies such as smart test selection, parallel test execution, and intelligent caching, teams can significantly accelerate build times without compromising code quality. Unit tests provide immediate feedback, enabling developers to catch regressions early and refactor confidently, although they need to be complemented by integration and end-to-end tests for comprehensive coverage. The use of test doubles, parameterized testing, and snapshot testing enhances test reliability and maintenance. Despite the maintenance overhead and potential false confidence they may introduce, unit tests are foundational in preventing bugs, encouraging modular design, and supporting test-driven development (TDD). Modern platforms like Harness CI leverage AI-powered test intelligence and automated tools to optimize test cycles and infrastructure use, demonstrating the potential for efficient, scalable testing within CI/CD environments.

Harness has developed a new Filters component system in React to enhance data filtering for improved developer productivity, addressing the limitations of their legacy system, which suffered from poor discoverability, inconsistent behavior, and high developer overhead. The new system, built through several iterations, utilizes the React Context API to provide a reusable, type-safe, and efficient framework that decouples filters from the framework internals, facilitates URL synchronization, and incorporates the principle of Inversion of Control (IoC). This design allows developers to define what their code does while the framework orchestrates when it happens, leading to a flexible and scalable filtering experience that is easy to maintain and enhances user experience by enabling seamless updates, shareability, and bookmarkability. The architecture serves as a blueprint for managing complex UI state management and reusable component design, showcasing how clear objectives and architectural principles can create elegant solutions in modern frontend systems.

Argo CD, a Kubernetes-native continuous delivery controller, is designed for enterprises to standardize GitOps—a model where Git serves as the source of truth, enabling a pull-based reconciliation loop that ensures cluster states align with Git declarations. This guide describes installing Argo CD using Helm for version control and predictable upgrades, emphasizing secure access with SSO, RBAC, and multi-team AppProjects for safe GitOps adoption. It outlines installing Argo CD as a shared infrastructure with high availability, monitoring, and backup plans to maintain reliability at scale. The installation process involves choosing between Helm for repeatable, audit-friendly setups or upstream manifests for quick evaluations, and it highlights the importance of stable operations through version pinning, ingress configuration, and TLS management. Enterprise practices stress the significance of monitoring, backup strategies, and controlled upgrades, treating Argo CD as a foundational platform infrastructure rather than a mere tool, with a focus on high availability and operational maturity for shared dependency management.

NoSQL databases, with their flexibility, scalability, and high throughput, have become essential components in modern application architectures, particularly as organizations transition to microservices and distributed systems. However, this flexibility can lead to operational, governance, and compliance risks without structured change control. Integrating NoSQL change management into CI/CD pipelines is crucial for maintaining deployment safety, governance, and traceability, allowing for automated versioning, testing, and rollbacks to mitigate compliance risks. Harness Database DevOps extends CI/CD best practices to NoSQL, providing automated delivery, governance, and observability, which enable enterprises to innovate rapidly without sacrificing reliability or audit readiness. By treating database changes as first-class software artifacts, enterprises can achieve greater deployment confidence, sustain release velocity, and enhance alignment across development and operations teams. Animesh Pathak, a Developer Relations Engineer at Harness, emphasizes the importance of automated governance and CI/CD integration to ensure NoSQL deployments remain safe, compliant, and resilient, while supporting high-velocity software delivery.

Aditya Kashyap discusses the evolution from traditional automation in DevOps to the emerging concept of Agentic AI, which represents a significant paradigm shift in operational philosophy. Unlike conventional automation, which is deterministic and brittle, Agentic AI offers a more dynamic approach by running scripts, detecting errors, analyzing logs, and applying fixes autonomously, effectively creating a "synthetic brain" for operations. This transition is framed within the C-P-A Model (Context, Planning, Action), transforming large language models into decision engines through perception, memory, reasoning, and tool use. Agentic AI enables the creation of "Living Architecture," ensuring infrastructure diagrams remain accurate by detecting and correcting deviations in real-time, and supports intent-based provisioning for optimized resource management. As DevOps evolves, the role of engineers shifts from scriptwriters to designers of Cognitive Architectures, managing AI-driven systems that autonomously handle tasks, improve governance, and speed up delivery processes. This transformation will unfold across three horizons, gradually increasing the autonomy of AI agents, culminating in a future where humans oversee policies and goals while AI manages routine operations. Kashyap emphasizes that rather than replacing engineers, AI will elevate their roles, allowing them to focus on strategic design and oversight.

In Q4 2025, Harness released significant updates to its Continuous Delivery, GitOps, and Continuous Verification platforms, focusing on enhancing deployment safety, infrastructure integration, and scalable workflows. The updates include support for deploying Google Cloud Managed Instance Groups (MIGs) and multi-account AWS CDK deployments, automated ECS blue-green traffic shifting, and improved Azure WebApp API rate-limit resiliency. Harness also integrated its Artifact Registry as a native source for various deployment types and enhanced Terraform steps and AWS connector session durations. New features in Continuous Verification allow for custom webhook notifications and cross-project GCP Operations health sources. The pipeline updates aim to reduce operational blind spots and improve execution monitoring, with enhancements like pipeline notifications, Bitbucket Cloud Connector support, dynamic input set branch resolution, and fault-tolerant trigger processing. GitOps improvements introduce ApplicationSets as first-class entities for streamlined application management and support for Argo Rollouts for advanced delivery strategies. These updates collectively aim to make releases safer, reduce operational overhead, and enable scalable delivery without adding complexity.

Backstage, originally developed by Spotify, is an open-source framework for internal developer portals (IDPs) that addresses challenges in platform engineering, such as onboarding and documentation management. However, maintaining and scaling Backstage requires significant engineering resources, leading many teams to explore alternatives. These alternatives typically fall into three categories: building and maintaining a self-hosted Backstage, opting for a managed Backstage service, or purchasing a commercial IDP like Harness. Commercial IDPs offer advantages such as faster implementation, built-in governance, and integration with existing CI/CD and infrastructure as code (IaC) systems, but require some trade-offs in customization and potential vendor lock-in. The decision between these options depends on factors like engineering capacity, governance needs, and desired level of control over the portal's functionality.

The transition from traditional DevOps practices to Platform Engineering is marked by the adoption of a "Golden Path" approach, which aims to streamline and standardize the path to production while embedding governance and security into every stage of the development pipeline. This concept emphasizes reducing cognitive load for developers and ensuring that compliance and security are integral parts of the workflow rather than afterthoughts. The architecture of a Golden Standard Pipeline is not tool-dependent but is defined by its layers of validation, including governance, security orchestration, supply chain security, and immutable delivery. Key principles include implementing governance as the initial step to prevent non-compliant processes, parallelizing security checks to enhance efficiency, ensuring supply chain integrity with metadata like SBOM and cryptographic signing, and maintaining artifact immutability to avoid environment-specific rebuilds. By decoupling policy from pipeline and investing in a "shift left" culture, organizations can future-proof their platforms and position the Golden Pipeline as a valuable product that enhances developer experience while maintaining high standards of trust and security.

Modern application security is challenged by the rapid pace of AI-driven development and cloud-native architectures, which increase both microservices and pipelines, often leaving DevOps teams responsible for catching vulnerabilities before production. Traditional methods of bolting security onto CI/CD pipelines are insufficient, causing alert fatigue and slowing down processes due to numerous false positives. Harness introduces AI-powered application security testing natively within its platform to reduce noise and improve trust in security findings by focusing on vulnerabilities that are actually reachable in production code. This pipeline-native approach allows for scalable security testing with pre-configured, reusable steps, enhancing operational efficiency by integrating security as a seamless part of the software delivery process. The Harness platform's integration of SAST and SCA tools, originally from Qwiet AI, provides deep visibility into open-source risks and offers AI-powered remediation suggestions, facilitating faster and more reliable vulnerability management. The visual workflow in Harness's STO simplifies integration, allowing developers to drag and drop security steps into their pipelines without complex configurations, thus improving security coverage and reducing the operational burden typically associated with application security testing.