Transforming Engineering Leadership with AI and Metrics

GitHub Universe 2025 & Engineering Intelligence: A Deep Dive with Old Stacks

Key Concepts:

• Engineering Intelligence: Utilizing data and analytics to optimize software delivery and maximize the return on investment in engineering teams.
• Value Stream Management: Analyzing the entire software delivery process from ideation to production to identify bottlenecks and improve flow.
• DORA Metrics: Key metrics (Deployment Frequency, Lead Time for Changes, Change Failure Rate, Time to Restore Service) used to assess software delivery performance.
• Deep Research Agents: AI-powered tools designed to analyze complex data sets, identify risks, and provide actionable insights for software delivery.
• Watermelon Projects: Projects that appear green (on track) on the surface but are actually facing significant issues internally.
• Context Switching: The act of rapidly switching between different tasks, which can negatively impact developer productivity and happiness.
• Flow Metrics: Metrics focused on the flow of work through the software delivery pipeline, such as cycle time and throughput.

I. GitHub Universe 2025: A Celebration of Community & Innovation

GitHub Universe 2025 was highlighted as a key event for the developer community, fostering collaboration and showcasing the latest advancements in software development. The event saw significant growth, with one new developer joining the platform every second. A central theme was the integration of AI, with GitHub emphasizing the power of AI agents working alongside developers. The event’s atmosphere was described as energetic and forward-looking, with a focus on “unblocking the path so you can build what’s next.” The next Universe event is scheduled for October 28th and 29th, 2026, at Fort Mason Center.

II. The Challenge of Measuring Engineering Value: Introducing All Stacks

The conversation with Jeff Kais, Field CTO at All Stacks, centered on the challenges of effectively measuring and maximizing the value of engineering investments. Kais, with over 30 years of experience in tech, explained that All Stacks addresses the problem of visibility into software delivery. Unlike manufacturing, the work of developers is often opaque, making it difficult to understand what’s being produced and how efficiently.

All Stacks provides metrics and insights across the entire delivery pipeline, including:

• Delivery Metrics: Tracking the speed and reliability of software releases.
• Value Stream Flow: Analyzing the flow of work from ideation to production.
• DORA Metrics: Monitoring key performance indicators related to software delivery performance.
• Developer Experience: Assessing the factors that impact developer productivity and satisfaction.
• R&D Capitalization & Automation: Measuring the efficiency of R&D spending and automation efforts.

The platform aims to transform these metrics into actionable insights for engineering leadership, enabling better forecasting and team management. All Stacks is designed to be valuable for teams of all sizes, starting at around 100 developers.

III. The "Watermelon Project" Problem & the Need for Deeper Insights

Kais illustrated a common problem in software development – the “watermelon project.” These projects appear green (on track) on the surface, often through optimistic reporting, but are actually facing significant internal issues. This disconnect highlights the need for deeper, data-driven insights beyond superficial project status updates. The traditional method of relying on project managers to gather status updates through interruptions is inefficient and prone to inaccuracies.

IV. Introducing All Stacks’ Deep Research Agents: AI-Powered Intelligence

All Stacks announced the launch of a new suite of AI-powered “Deep Research Agents” designed to address the limitations of traditional dashboards and provide more actionable intelligence. These agents are described as “multi-prompt, iterative” and analyze data across the entire software delivery pipeline.

Key features of the Deep Research Agents include:

• Delivery Risk Deep Research Agent: Analyzes projects, code, track record, code quality, and PR review cycles to assess the likelihood of on-time delivery and the quality of the delivered code. It then provides a list of actions to mitigate identified risks, including assigning tasks to specific team members.
• Workflow Analysis Agent: Identifies bottlenecks and inefficiencies in the software delivery workflow by analyzing different states of the workflow, happy paths, and areas where code gets stuck.
• Requirements Readiness Agent: Scans requirements documentation to identify gaps and suggest missing elements, such as multi-factor authentication.

These agents aim to move beyond simply presenting data to proactively identifying problems and suggesting solutions.

V. The Importance of Context & Avoiding Metric Misinterpretation

Kais emphasized the importance of interpreting metrics correctly and avoiding the pitfalls of focusing solely on vanity metrics. He cautioned against blindly chasing metrics like lines of code or deployment frequency without considering the broader context of the software delivery pipeline. He highlighted that simply increasing code output doesn’t necessarily translate to increased value if other parts of the pipeline (CI/CD, testing, deployments) are not keeping pace. Recent research indicates that CI/CD pipelines are actually slowing down despite increased code output.

VI. Shifting the Focus: From Output to Investment & Alignment

Kais stressed the need to communicate the value of engineering work in terms of return on investment (ROI) to business stakeholders. He recommended focusing on two key metrics:

• Investment Maximization: Demonstrating how engineering efforts are contributing to key business initiatives.
• Alignment: Proving that engineering teams are focused on the right things, avoiding the trap of cognitive overload and excessive context switching.

He argued that engineering leaders should be able to demonstrate how their work directly impacts business outcomes, rather than simply reporting on technical metrics.

VII. Developer Experience & the Power of Context Switching Analysis

The discussion also touched on developer experience (DevEx). Kais argued that measuring developer happiness through surveys is often unreliable. He proposed focusing on context switching as a more objective indicator of developer productivity and well-being. By analyzing where developers spend their time, organizations can identify and address sources of interruption and distraction, allowing developers to focus on deep work and deliver higher-quality code. All Stacks provides time sheet-level fidelity data to track developer activity and identify patterns of context switching.

VIII. Future Roadmap & the Orchestrated World of AI-Assisted Development

Looking ahead, All Stacks plans to continue developing AI-powered tools to help organizations navigate the evolving landscape of software development. The company anticipates a future where AI plays an increasingly important role in automating tasks, identifying risks, and optimizing the entire software delivery pipeline. The focus will remain on helping organizations maximize their engineering investments and deliver value faster.

Where to Learn More:

• All Stacks: https://www.allstacks.com/
• Flowtopia: https://flowtopia.io/
• Jeff Kais on LinkedIn: Jeff Dais (profile available on LinkedIn)