November 2024 Summaries

Bluesky is a rapidly growing social network that provides users with flexibility and control over their content, including highly configurable moderation tools and customizable feeds using the open source ATProtocol. One potential issue for the platform is the rise of vaccine disinformation, which could be mitigated by using large language models (LLMs) to filter posts containing misinformation or denialism. LaunchDarkly's feature flags can help decouple deployment from feature management on platforms like Bluesky that are scaling quickly. In this tutorial, a custom Python-based Bluesky feed is created with keyword-based filtering and LLM sentiment analysis, which can be conditionally enabled using a LaunchDarkly feature flag. Future upgrades could include more advanced machine learning techniques for detecting vaccine-related content and utilizing AI configs to improve sentiment analysis.

LaunchDarkly introduces a new, simplified metric creation flow for setting up metrics that align with user goals. The updated process is more intuitive and user-friendly, featuring a type-ahead experience to select the right event key faster. Key benefits include simplicity, speed, accuracy, and confidence in metric setup. Users can track various impactful metrics such as conversion rate, engagement frequency, performance metrics, and revenue impact to measure the success of their feature releases and experimentation efforts.

Companies are under immense pressure to innovate quickly, but this drive for speed can come at the expense of software stability. In tech, the mantra "Move fast and break things" is well understood, however, in today's landscape, breaking things isn't necessary. Effective risk management practices are vital for sustainable innovation. Tech leaders need to weave smart risk management practices into their development workflow, such as progressive rollouts, automated monitoring tools, instant rollbacks, targeted segmentation and runtime configuration management. A cultural shift is also required, with a company culture that prioritizes both speed and preparedness being fundamental to sustained and scalable innovation. By fostering an environment where innovation is balanced with operational excellence, leaders can ensure their organizations "move fast" and thrive in the digital era, with no breaking required.

The text discusses the importance of experimentation in decision making, drawing inspiration from the story of MIT students who used small experiments to count cards and beat casino odds. It emphasizes that using high-quality data for comparison between control and treatment can significantly improve decision accuracy. The article suggests aiming for incremental improvements rather than large changes, as they are easier to implement and less risky. A culture of experimentation is encouraged, with a focus on making more, better decisions based on causal data. It also highlights the value of velocity in small wins that add up to significant overall business impact. The text concludes by recommending leveraging an experimentation program to mitigate risks associated with larger projects and using learnings from smaller experiments to inform decision-making on bigger features.

This tutorial teaches how to use LaunchDarkly feature flags with a Contentful + React web application. It covers the basics of using a headless CMS like Contentful and integrating it with React for rendering content. The tutorial also demonstrates adding a LaunchDarkly feature flag to control the display of product tags in the catalog, allowing developers to personalize the website for specific audiences or run A/B experiments.

LaunchDarkly has revamped its Developer training program with an interactive, self-service workshop and shorter video lessons to improve software release management skills. The changes were made based on user feedback, which highlighted the need for more hands-on components and real-world examples. The updated training provides a balanced approach that combines theoretical understanding with practical application. In 2025, LaunchDarkly Academy plans to expand its offerings with additional courses, certifications, and hands-on labs to support users at every stage of their software release management journey.

LaunchDarkly has introduced several new features to improve feature management at enterprise scale. These include the Launch Insights Dashboard for assessing release practices, automated flag cleanup checks, Custom Role Builder for creating custom roles and policies, test run flag targeting rules, and updated ServiceNow integration for automatic approval and application of changes in LaunchDarkly when reflected in ServiceNow. These updates aim to enhance efficient, scalable, and effective feature management across organizations of all sizes.

In 2024, software development teams faced challenges with poorly executed rollouts and failed app launches. LaunchDarkly introduced Guarded Releases to address these issues by ensuring every release is monitored, allowing for quick rollbacks when problems arise, and enabling phased rollouts. The new features also provide insights into release health, real-time alerts, enhanced usability, and error monitoring. Additionally, LaunchDarkly introduced AI configs to manage the configuration of prompts, models, and configurations that power GenAI applications at runtime. New ways to manage releases at scale include a Launch Insights dashboard, flag cleanup made easy, custom role builder, and updated ServiceNow integration.

The rapid evolution of LLMs and GenAI presents numerous opportunities for software teams across industries, but also introduces unique challenges when bringing these non-deterministic applications to production at an enterprise scale. Key challenges include managing hundreds of prompts in code, keeping up with the rapid pace of new models, measuring the impact of changes, and mitigating the unknown risks of non-deterministic models. LaunchDarkly's AI configs offer a modern tool for managing GenAI products at runtime, allowing teams to create and manage prompt and model configurations, instantly roll back to stable configurations in case of issues, visualize and monitor key metrics like token usage, target AI configs to various audiences, and run multiple configurations against the same application. This helps accelerate GenAI development and bring safer, more reliable AI features to users.

LaunchDarkly has implemented a solution called Guarded Releases to improve their release monitoring process. This feature allows for automatic rollbacks triggered by real-time error thresholds, providing actionable insights between flags and errors. The Fundamentals Engineering Team at LaunchDarkly works closely with product engineers to develop impactful features and products faster, benefiting both internal teams and customers. Guarded Releases have been successful in minimizing the impact of software changes on end users and allowing for quick identification and resolution of issues.

Software bugs often arise from new code changes, leading to costly outages and customer dissatisfaction. To address this issue, LaunchDarkly has introduced Guarded Releases, a solution that equips teams with proactive safeguards for fast, resilient updates while minimizing risk. By incorporating automated monitoring, controls, and rollbacks into the deployment process, teams can prevent issues before they impact users, making each release low-risk and high-confidence. Guarded Releases enable developers to maintain a balance between speed and reliability in software releases, ultimately enhancing customer experiences and driving productivity.

LaunchDarkly has integrated with AWS PrivateLink, allowing mutual customers to reduce data transfer costs and improve security by keeping their traffic within the AWS network. This integration is particularly beneficial for users of LaunchDarkly's data-intensive product offerings, as it can significantly lower event data egress costs and enforce stricter firewalls. The integration works by providing an endpoint service that enables direct data transfer from AWS to LaunchDarkly without using a regular internet connection.

LaunchDarkly introduces automated checks to help manage feature flag sprawl in high-velocity organizations. The proliferation of stale flags can clutter the codebase and disrupt developer workflows, leading to tech debt. Automated archival checks mark flags as Needs code removal or Ready for Archival based on specific criteria. This helps teams more easily manage flag debt, contributing to a cleaner codebase and freeing up time for building new products. In the future, users will be able to customize these definitions to meet their organization's needs.

This tutorial demonstrates how to use LaunchDarkly feature flags to toggle between two different email providers (Resend and Mailgun) in an ExpressJS application. The process involves setting up a development environment with Node.js, git, and npm installed, creating accounts for Resend, Mailgun, and LaunchDarkly, and then building the example ExpressJS application. By using feature flags, developers can more safely migrate from one email provider to another or use multiple providers as primary and backup services.

Release management is a crucial process in software development that involves planning, scheduling, and controlling the delivery lifecycle from development to deployment and beyond. It ensures efficient delivery of new features, bug fixes, and improvements with minimal disruption. Modern release management practices emphasize continuous integration and continuous delivery (CI/CD) for more frequent, smaller releases. Key stages in the software development process include planning, design, development, testing, deployment, testing in production, release, and maintenance. Effective release management accelerates time-to-market, enhances software quality, manages risks, improves collaboration, increases visibility and predictability, optimizes resources, supports scalability, boosts customer satisfaction, and enables continuous improvement. Strategies to achieve better release management include adopting automation, using feature flags, implementing better testing strategies, improving monitoring and observability, streamlining communication and collaboration, standardizing and documenting processes, focusing on continuous improvement, implementing progressive delivery, leveraging cloud-native technologies, and investing in the right tools.

Deploying applications on Kubernetes can be challenging due to the need to manage containers, pods, services, and configurations. However, employing the right Kubernetes deployment strategy can help minimize risk when rolling out new features, maintain service availability during updates, better manage resources across your cluster, quickly roll back if something goes wrong, and test new versions of the application with real traffic before full release. There are six types of Kubernetes deployment strategies: recreate deployment, rolling update, blue/green deployment, canary deployment, shadow deployment, and A/B testing deployment. Each strategy provides different tradeoffs between speed, safety, complexity, and resource usage. Combining your deployment strategy with feature flags gives you an extra layer of control and safety by controlling what happens once the code is in production.