September 2025 Summaries

Weave, a software engineering intelligence platform co-founded by Andrew, is designed to measure and optimize the use of AI in code editors, agents, and review tools by integrating large language models (LLMs) and domain-specific machine learning. Initially launched to measure engineering productivity, Weave now offers two suites of products: one for optimizing AI setup and another for optimizing AI usage. The platform provides tools like a Cursor rules linter, which analyzes and refines rules files for AI coding tools, and a Weave Score metric to track the productivity impact of AI decisions. Additionally, Weave includes detection logic to identify AI contributions in code repositories and offers insights into the financial savings from AI adoption by correlating AI-generated work with user wage data. Through these innovations, Weave aims to help engineering teams improve their efficiency and better understand when and how to effectively utilize AI.

In recent interviews with 134 engineers, a significant observation emerged: the untapped potential of junior engineers, who are often overlooked by companies in favor of senior talent. While businesses grapple with a scarcity of senior engineers and the high costs associated with them, they miss out on the fresh energy, adaptability, and long-term potential that motivated juniors bring to the table. Many companies avoid hiring juniors due to perceived onboarding challenges, outdated interview processes, and misconceptions about productivity. However, juniors, especially those who are AI-native, can quickly become productive with appropriate mentoring and the use of modern tools. Companies like Shopify are capitalizing on this by hiring interns and juniors, recognizing the competitive edge they offer. Investing in junior talent, with a focus on mentoring and leveraging AI, can create a robust pipeline of future senior engineers and reduce dependency on an expensive and limited senior talent pool.

The post-COVID-19 work culture has seen an increase in meetings, which has negatively impacted software engineers' ability to engage in "deep work," a concept popularized by Cal Newport that involves focused, undistracted work leading to high productivity and skill improvement. The discrepancy between managers' and engineers' perceptions of meetings exacerbates this problem, with managers often assuming that AI coding tools allow engineers to work effectively in smaller time segments. However, this assumption undermines the flow state necessary for deep work, as frequent context switching and multitasking during meetings reduce productivity and work quality. The article argues that effective meeting management and scheduling, as well as fostering a culture that prioritizes deep work, are crucial for maintaining high performance among engineering teams. It highlights the example of Pylon's engineering team, which minimizes meetings and code reviews to enhance productivity, suggesting that providing substantial uninterrupted time can help engineers achieve better results in a future increasingly reliant on AI.

AI-powered engineering analytics, as exemplified by platforms like Weave, are revolutionizing the developer experience by providing insights that go beyond traditional metrics, such as lines of code or commit counts, which are poor indicators of productivity. Weave leverages large language models and domain-specific machine learning to analyze engineering work, offering visibility into code complexity, AI integration, and individual and team performance. This approach helps identify friction points in the development process, enhancing productivity and satisfaction by focusing on feedback loops, cognitive load, and flow state. By understanding the intricacies of engineering tasks and the real impact of AI tools, Weave enables teams to optimize processes and achieve better outcomes, with companies reporting significant improvements in output and developer retention.

Engineering teams face the dilemma of whether to build or buy an analytics platform to improve software delivery performance, with the decision hinging on factors like resources, timelines, and business goals. Custom-built solutions offer complete control and customization to meet specific needs, but they require significant time investment, ongoing maintenance, and dedicated resources, which can strain non-core business functions. In contrast, purchasing a ready-made platform like Weave allows teams to gain immediate insights, benefiting from proven expertise and ongoing vendor support, although integration and customization still demand some internal resources. Weave, leveraging advanced LLMs and machine learning, provides fast, actionable recommendations and seamless integration into existing workflows, making it an attractive option for teams seeking quick improvements without the complexities of traditional metrics frameworks. Many successful teams adopt a hybrid strategy, starting with a platform like Weave to gain initial insights before developing custom extensions to address specific needs, reflecting a broader industry trend towards bolder technology investments.

In today's rapidly evolving engineering landscape, measuring the return on investment (ROI) of AI-powered engineering tools has become crucial, yet challenging, due to the non-linear nature of engineering productivity. With only a small fraction of organizations effectively tracking developer productivity beyond basic metrics, AI-driven analytics platforms like Weave are revolutionizing this space by analyzing work patterns and revealing actionable insights. These tools help track team output, identify bottlenecks, and optimize resource allocation, thereby translating engineering improvements into tangible business outcomes such as revenue acceleration, cost reduction, and risk mitigation. Establishing baseline metrics, connecting them to business impacts, and addressing opportunity costs are essential steps in building a compelling business case for these tools. While traditional metrics may not fully capture the value of improved developer satisfaction and team dynamics, a comprehensive ROI analysis should combine quantitative savings with qualitative benefits to ensure successful adoption and stakeholder buy-in.

AI tools are increasingly integral to engineering workflows, with 76% of developers incorporating them according to Stack Overflow's 2024 Developer Survey, yet measuring their true impact presents challenges. Traditional metrics like tool usage frequency and licenses purchased fail to capture the effectiveness of AI tools in enhancing productivity, code quality, and problem-solving capabilities. A more effective approach involves measuring output-based metrics such as code velocity, bug reduction, and documentation quality, alongside context-aware usage patterns and qualitative feedback to assess the real benefits and drawbacks of AI assistance. Tools like Weave can connect AI usage to team performance, while platforms like SonarQube and CodeClimate help track code quality, emphasizing the need for a holistic strategy to understand AI's role in development. By focusing on meaningful outcomes rather than mere activity, teams can optimize their AI usage and gain a competitive edge in the rapidly evolving AI-driven landscape.

Choosing between building or buying a software development metrics platform is a critical decision for engineering leaders seeking better visibility into team performance. Building a custom platform offers a tailored fit and avoids vendor lock-in but requires significant time, expertise, and ongoing maintenance. For larger teams, building may be financially viable, yet it involves a complex development process lasting one to two years. Conversely, buying a commercial solution provides quick insights and leverages refined platforms with broad feedback, allowing teams to focus on core business activities rather than internal tooling. Small teams are generally advised to purchase due to lower overhead, while medium-sized teams must weigh their engineering culture and resources. Large teams might consider building only if they have the resources for a long-term commitment. Hybrid approaches can combine the immediate benefits of commercial solutions with custom extensions. Ultimately, unless unique requirements or compliance demands necessitate a custom build, buying often offers a faster return on investment and allows organizations to concentrate on their primary objectives.