Skip to content

AI Era Development Culture and Methodology: Evolution from Agile to Hybrid Development

Introduction

The world of software development has undergone a major transition from waterfall to agile development, and is now facing another wave of transformation driven by the rapid advancement of AI technology. Research conducted in 2024-2025 has revealed that a new paradigm beyond traditional development methodologies is being established.

This article provides a comprehensive explanation of changes in development culture and methodologies in the AI era through the latest research findings, corporate case studies, and practical insights.

The Current State of AI-Driven Development: 2024-2025 Reality

Definition of AI-Driven Agile Development

AI-driven agile development is defined as a methodology that actively utilizes generative AI and AI agents in each phase of the agile development process, automating and enhancing everything from implementation, testing, and progress management to market and user data analysis and scenario simulation.

Actual Implementation Effects and Challenges

Success Case: SmartHR (2025) - Productivity increased by more than 30x through human-AI collaborative teams - Activated dialogue with stakeholders by utilizing AI in sprint review preparation

However, interestingly, a 2025 empirical study of 246 tasks by 16 experienced developers reported an unexpected result: task completion time increased by 19% when using AI. On the other hand, the DORA Report 2024 shows measured data indicating approximately 2.1% productivity improvement in companies that have adopted AI2, suggesting the importance of proficiency and application scenarios in AI utilization.

The Era of Small Batch Development

A future scenario depicts AI agents ("Codius," "Imagine," "Sentry," "Optima," etc.) sharing task progress and system status with each other in real-time 24/7 while working in small batches measured in minutes and hours. This is a continuous delivery approach based on Trunk-Based Development (TB development) principles3.

Changing Developer Roles: From Coder to Orchestrator

Fundamental Role Transformation

High-growth SaaS companies in 2025 report a dramatic change: 90% of code is AI-generated (compared to 10-15% twelve months earlier). Accordingly, developer roles have changed as follows:

Previously: Code creator Currently: Code manager, editor, orchestrator

Newly Required Skill Sets

Technical Skills

  • Basic understanding of AI and data science
  • Quality control and correction ability for AI-generated code
  • Adaptability to multiple languages and frameworks
  • System architecture design capability

Human Skills (becoming more important)

  • Interpersonal relationship building (most important skill)
  • Creativity and problem-solving ability
  • "Question framing ability" and "hypothesis formulation/verification ability"
  • Vision formulation and strategic thinking capability

Emergence of New Roles

  • AI Developer Advocate
  • Automation Lead
  • Specialists managing human-AI collaboration

According to a report by Japan's Ministry of Economy, Trade and Industry, "tasks" are being significantly reduced, changing the role of people, including specialized personnel, to more creative ones, making uniquely human creative skills (entrepreneurial spirit, etc.) and business/design skills more important4.

Changes in Team Composition

With the evolution of AI tools, the author estimates that by 2026, a team of 5 people could achieve the same results as a 15-person team in 2020, with expectations for a shift toward smaller, more elite team compositions.

Evolution of Agile Development: Outlook Toward 2030

Elevation to Management Methodology

Agile development is transcending the realm of "software development methodology" and being treated as a "management methodology for developing and improving business services."

Example: Amazon - Operated approximately 3,300 two-pizza teams (small teams) in 2018-20191 - Practiced agile working methods company-wide - Developed their unique "Working Backwards" process

Human-AI Bromance Era

A new collaborative relationship called "Human-AI Bromance" is predicted to be established toward 2030:

  • Symbiotic collaborative relationship between humans and AI systems
  • Mutually complementary partnership leveraging each other's strengths
  • Achieving efficiency, innovation, and customer satisfaction while maintaining human-centric values

Rediscovery of Technical Excellence

Over the past decade, speed-focused development sacrificing quality has been common, but the following recognition is now spreading:

  • Solid technical practices are essential for sustainable agility
  • Extreme Programming (XP) practices are being reevaluated
  • Achieving both quality and speed is becoming more important

Reevaluation of Requirements Definition and Design in Large-Scale Development

Why Requirements Definition Has Become Important Again

The reality that "the reason system development fails is requirements definition, now as in the past" remains unchanged. Particularly in AI development:

  • The quality of requirements definition determines the direction of AI coding
  • Large-scale development with ambiguous instructions is high-risk
  • The cost of changing logic/architecture midway is enormous

Amazon Kiro's Innovative Approach

Amazon's Kiro, announced in July 2025, adopts a specification-driven development approach5:

  • Automates requirements analysis, design, and task creation
  • Specializes in project planning, specifications, and code documentation
  • While Amazon Q Developer emphasizes code completion, Kiro emphasizes the design phase

Realistic Challenges in Large-Scale Development

Limitations of the "immediately build something that works" approach:

  • Logic breaks down midway, frequently requiring complete rebuilding
  • Failures due to insufficient consideration of constraints at the architecture design stage
  • Non-functional requirements (performance, security, maintainability) are insufficient with implementation tweaks alone

Important Perspective: - Small-scale development: "working prototype first" agile is effective - Medium to large-scale development: "foundation building" is essential

Rise of Hybrid Development Methodologies

Big Tech Company Practices

Facebook, Apple, Amazon, Netflix, Google, and Microsoft have not widely adopted traditional agile frameworks, instead developing their own unique methodologies:

  • Amazon: Working Backwards process
  • Basecamp: Shape Up methodology
  • Google: Data-driven iterative approach with Trunk-Based Development

Realistic Hybrid Approach

Actual companies practice the following divisions:

  • Upstream processes (requirements definition, basic design): Phase to clarify specifications (not frozen but iteratively updated)
  • Midstream processes (detailed design, coding, unit testing): Agile iterative development
  • Downstream processes (integration testing, comprehensive testing): Systematic quality assurance phase

Unlike traditional waterfall, this is a flexible approach with the premise that specifications are continuously updated.

Agile vs. Waterfall in the AI Era

Academic research has revealed the following:

Suitability for AI/ML Projects - Agile: Naturally fits the experimental nature of AI development - Waterfall: Still effective in areas with clear requirements and few changes

Important Decision Criteria The presence or absence of specification changes is the most important factor: - No specification changes → Waterfall - Frequent specification changes → Agile

Practical Cases: Corporate AI Development Culture Adoption

Japanese Company Success Cases

GMO Internet Group - Achieved 670,000 hours of operational efficiency through AI implementation - Established company-wide AI utilization culture

Major Bank - Reduced over 220,000 hours of monthly effort through ChatGPT utilization - Planned approximately 50 billion yen AI investment through 2027

Toyota - Developing "Mobility AI Platform" with NTT - Planning 500 billion yen investment through 2030

Netflix - Small cross-functional teams - Sprint-centered iterative development - Continuous improvement through A/B testing and chaos engineering

Microsoft - Focusing on AI agent creation tools for enterprise customers - Agent-to-agent collaboration and data access control features

Characteristics of AI Era Development Culture

Transparency and Explainability

Important elements in human-AI collaboration: - Transparent reasoning process: Trackable AI thinking process - Multi-turn conversations: Maintaining context across entire codebase - Continuous feedback loop: Ongoing improvement of AI output

Integration of Security and Quality

  • Vulnerability scanning of AI-generated code
  • Real-time quality checking
  • Automation of secure development practices

Global Borderless Development

  • Cross-border collaboration through real-time translation tools
  • Realization of 24-hour development cycles
  • Innovation creation leveraging diversity

Predictions Toward 2030

Technological Evolution - Exponential improvement in processing power through quantum computing - Dramatic shortening of development cycles - Collaboration with more advanced AI agents

Organizational Changes - Return to agile principles and values - Emphasis on simplicity and customer value delivery over complex frameworks - Balance of utilizing AI while maintaining human-centric values

Practical Recommendations

Organizational Level

  1. Gradual AI adoption: Expand sequentially from small success experiences
  2. Human resource investment: Education and training for AI collaboration skills
  3. Cultural transformation: Organizational culture treating failures as learning opportunities

Team Level

  1. Adopt hybrid approach: Methodology selection according to project characteristics
  2. Continuous learning: Response to rapid AI technology advancement
  3. Human-centered design: Position AI as a partner, not a threat

Individual Level

  1. Strengthen core skills: Creativity, problem-solving, communication
  2. Acquire AI collaboration skills: Ability to evaluate and improve AI output
  3. Vision building ability: Ability to bridge technology and business

Conclusion

AI era development culture and methodologies are evolving beyond the traditional agile vs. waterfall dichotomy toward more flexible and adaptive hybrid approaches.

What's important is not viewing AI as merely a tool, but integrating it as a member of the development team and building a new collaborative relationship combining human creativity and AI efficiency.

For development organizations to succeed from 2025 onward:

  • Reinvestment in requirements definition and design quality
  • Maintenance of human-centric values
  • Strengthening continuous learning and adaptability
  • Establishment of AI collaboration culture

Establishing a new development culture integrating these elements will be essential.

Rather than fearing change, viewing change as an opportunity and actively exploring new development paradigms in the AI era will be the source of sustainable competitive advantage.

References

  1. AWS materials and lecture records (2018-2019) "Achieving Business Agility with Scrum@Scale" and others ↩

  2. Google Cloud "DORA Report 2024" (2024) ↩

  3. Trunk-Based Development official site (https://trunkbaseddevelopment.com/) ↩

  4. Ministry of Economy, Trade and Industry "Human Resource Development in the Digital/Generative AI Era" (2024) ↩

  5. AWS News Blog "Kiro: The AI IDE for prototype to production" (July 2025) ↩