Developing with AI: Balancing Speed and Code Quality | Sonar Summit 2026

The Growing Challenge of Architectural Debt

Olivier Gdan, co-founder of Sonar, presented the newly released architecture capability in SonarQube at the Sonar Summit 2026, addressing a critical challenge facing modern software development. While organizations have successfully managed technical debt at the code level through tools like SonarQube for nearly two decades, focusing on maintainability, reliability, and security, a more insidious problem has emerged: structural debt. This architectural debt manifests when applications lack proper organization, unclear abstractions, and components that don't adhere to single responsibility principles. The consequences are tangible—engineers leave due to messy codebases, changes create difficult-to-trace side effects, and overall development velocity declines significantly.

Understanding the Nature of Architectural Erosion

The critical insight Gdan emphasized is that architectural problems don't appear suddenly but develop gradually through erosion. During initial application development, teams typically organize their code thoughtfully. However, as shortcuts accumulate to save time or due to insufficient upfront investment, the architecture gradually degrades. By the time teams feel the pain of poor architecture, remediation requires major structural changes. This preventive insight underscores why maintaining healthy architecture—characterized by clear understanding, modularity, and well-defined organization—is essential for sustaining long-term development velocity.

A Three-Pillar Approach to Architecture Management

Sonar's solution addresses architecture through three foundational concepts. First, current architecture represents the actual state of an application as discovered through code analysis at every build, providing a reverse-engineered map of existing components and their relationships. Second, intended architecture defines the target state—whether rooted in original design intentions or current strategic goals—creating a blueprint for stakeholders. Third, deviations identify gaps between current and intended states, forming the basis for actionable remediation. The product philosophy prioritizes making architecture comprehensible to entire teams, including both human developers and AI agents, rather than serving only specialized architects.

Implementation Through Four Strategic Pillars

The solution is built on four operational pillars designed for comprehensive architecture governance. The first enables all stakeholders—human and AI—to understand current architecture through accessible, live documentation that reveals structure, relationships, and component complexity. The second pillar simplifies the formalization of intended architecture, allowing teams to easily define and incrementally improve their architectural goals without mapping unnecessary complexity. The third pillar ensures clear prioritization of issues by preventing overlapping reports—when structural issues are identified, subsequent relation issues caused by those structural problems are suppressed to prevent duplicate reporting. Finally, the fourth pillar leverages SonarQube's existing capabilities—workflows, quality gates, and in-code issue visualization—to make architectural remediation actionable and integrated into development workflows.

Defining and Evolving Architecture Strategically

The intended architecture feature deliberately avoids mapping entire applications, recognizing that comprehensive mapping would require constant updates with every code change. Instead, teams identify and focus on what matters most, starting from top-level components and defining relationships between critical elements. This targeted approach creates a stable blueprint that serves as a reference point for all stakeholders, triggering warnings and quality gate failures when the actual architecture drifts from intended design. This flexible, incremental approach to architecture definition allows organizations to establish control over their most critical structural elements while maintaining agility in other areas.

Key Takeaways

• Architectural debt is a silent threat that erodes gradually through small shortcuts and accumulates until major refactoring becomes necessary; preventing drift requires continuous monitoring against intended architecture
• Architecture management spans three dimensions: component structure and hierarchy, inter-component relationships, and design specifications for how components should interact
• Intended architecture should be selective, not comprehensive, focusing on critical components and relationships rather than mapping entire applications to avoid constant maintenance overhead
• Effective solutions must be team-inclusive and actionable, supporting both human developers and AI agents while integrating with existing development tools and workflows
• Deviations should be prioritized intelligently by suppressing redundant issues and ensuring each architectural problem is reported once to maintain focus on root causes