Governing technical debt when the authors are agents

Thursday, June 25 at 5pm CEST | 10am CT

Managing technical debt has evolved from an internal hygiene issue into a critical reporting requirement. Boards now demand quantitative health metrics, and auditors require evidence of long-term maintainability. However, the emergence of the Agent Centric Development Cycle (AC/DC) has shifted the math. With 55% of software developers now regularly using AI agents, code is being drafted in massive, asynchronous payloads that outpace traditional human review.

This session walks engineering leaders through the strategic implications of agentic authorship. We will explore why detection alone is no longer a competitive advantage and how the next generation of code governance relies on verified code remediation within the software development workflow.

Best for: VPs and Directors of Engineering, platform engineering leaders, and architecture owners responsible for maintainability standards across a large surface area of services.

What you’ll take away

A multi-layer framing of code health: How debt accumulates differently in application logic, infrastructure as code (IaC), and the structural integrity of your system.
The Guide-Verify-Solve model: A blueprint for end-to-end management, from upstream prevention to automated, verified fixing.
Operational accountability: How agentic workflows change the requirements for code ownership, auditability, and the "trust and verify" model.
Architecture as active context: How to provide agents with engineering standards as an input—not a downstream filter—to avoid debt at the point of authorship.
Executive-ready visibility: A reference for cross-team dashboards that provide a single, defensible number for codebase health.

Agenda

The shift to AC/DC: Why faster authorship necessitates a new approach to code quality.
The four layers of debt: Application code, IaC, dependencies, and structural design.
The Guide-Verify-Solve cycle in practice: Moving validation into the agentic sandbox.
Verified remediation: What changes when a fix is validated by SonarQube Server or Cloud before it lands in the main branch.
Standards as input: Utilizing MCP servers to feed engineering norms to agents.
The Board-level view: Organization-wide reporting for technical leaders.
Live Q&A.

Tom Howlett

Tom Howlett is Director of AI Engineering Engagement at Sonar, leveraging over 30 years at the forefront of Software Development Practices. Starting as a developer in ‘90’s .com startups in London and NYC, he transitioned to guiding companies in adopting agile software engineering practices with a strong focus on quality. This commitment eventually brought him to Sonar, where he led Product Management before moving to AI Engineering Engagement. He now focuses on how developers can build an efficient Agent Centric Development Cycle without compromising on quality.