Moving away from Agile: What's Next – Martin Harrysson & Natasha Maniar, McKinsey & Company

Summary

This talk explores how AI is reshaping software development, highlighting the need for a paradigm shift in roles, workflows, and team structures. Martin and Natasha discuss how enterprises can harness AI to improve productivity and overcome bottlenecks by adapting their operating models. Key topics include AI-native workflows, evolving roles for developers and product managers, and the importance of change management and upskilling. The talk emphasizes the need for smaller, more dynamic teams and continuous planning to fully realize AI's potential, offering valuable insights on the future of AI in software development.

Takeaways

• 😀 AI is driving a paradigm shift in software development, similar to the transition to agile methodologies two decades ago.
• 😀 Despite impressive individual productivity improvements, organizations are still seeing only marginal gains at the company-wide level due to operational bottlenecks.
• 😀 Key bottlenecks include issues around team collaboration, manual code review, and the generation of tech debt with new AI-assisted code.
• 😀 The variability in AI tool effectiveness and user experience creates inefficiencies in task allocation and resource management for teams.
• 😀 A major bottleneck in AI-powered software development is how work is reviewed—manual reviews still dominate, even as AI accelerates code generation.
• 😀 Successful organizations are leveraging AI native workflows that scale across the entire software development life cycle, moving from point solutions to comprehensive integration.
• 😀 AI native roles are emerging, consolidating traditional engineering functions into smaller, more agile teams with cross-functional expertise in full-stack development.
• 😀 These AI-driven shifts lead to improvements in time to market, delivery speed, and overall software quality across teams.
• 😀 Continuous upskilling, measurement systems, and incentivization are critical to driving the adoption and effective use of AI tools in enterprise settings.
• 😀 Change management plays a pivotal role in scaling AI adoption, with a focus on clear communication, training, and performance tracking to ensure lasting impact.
• 😀 Enterprises that successfully integrate AI into their software development process are seeing economic benefits such as reduced labor costs, faster time-to-revenue, and enhanced feature quality.

Q & A

What is the main focus of the talk presented by Martin and Natasha?

The talk focuses on how AI is changing the software development process, particularly in terms of the operating models and the people involved in leveraging AI for software development.

What was a significant shift Martin experienced early in his career?

Martin recalls the shift to agile development methods nearly 20 years ago, where he worked with Kanban boards, stand-ups, and other ceremonies, which was a significant change for his company at the time.

What are the key bottlenecks that have emerged with AI in software development?

Some key bottlenecks include the lack of change in how teams collaborate, the manual review processes still being used despite faster code generation, and the amplification of tech debt due to automated code generation.

Why is AI adoption in enterprises often showing only marginal productivity improvements?

The marginal improvements are due to a disconnect between the potential of AI and its implementation in organizations. Issues like inefficient work allocation, uneven usage of AI tools, and manual review processes are limiting the impact AI can have on productivity.

How are modern software development teams changing to leverage AI more effectively?

Teams are reorganizing into smaller, more specialized pods with new roles. These roles focus on orchestrating AI tools rather than just executing traditional tasks like writing code or handling QA.

What role do AI-native workflows play in successful enterprises?

AI-native workflows enable organizations to scale AI usage across various stages of the software development lifecycle, increasing the effectiveness of AI tools by integrating them across multiple use cases instead of just isolated tasks.

What are AI-native roles, and why are they important?

AI-native roles involve creating smaller teams (pods) with individuals who can manage and orchestrate AI tools across the full software stack. These roles consolidate multiple engineering tasks, improving collaboration and efficiency.

What was the outcome of testing team-level interventions at a leading international bank?

The interventions led to a significant increase in the consumption of AI tools, a 51% rise in code merges, improved delivery speed, and better alignment with business priorities.

How does AI impact the role of product managers (PMs) in software development?

PMs are shifting from managing long product requirement documents (PRDs) to directly iterating on specifications with AI tools, enabling faster prototyping and closer alignment with customer feedback.

What does a successful change management strategy look like when scaling AI in software development?

A successful strategy involves ensuring clear communication, proper upskilling, incentivizing adoption, and setting up systems to measure the impact of AI tools. This helps address potential issues early on and ensures the changes are effective across the organization.

Why is a robust measurement system important for AI adoption in enterprises?

A robust measurement system helps track the outcomes of AI adoption, focusing on metrics like speed, productivity, code quality, developer satisfaction, and economic outcomes. This ensures that AI tools are driving real, measurable improvements rather than just increasing tool usage.