From Microservices to Agent-Services: Architectural Patterns for Autonomous System Boundaries

The rise of Agentic AI is introducing a paradigm shift in the microservices architecture: services are becoming autonomous agents that can redefine their own boundaries, merge capabilities with other services, or split responsibilities based on learned optimization patterns. That is why microservices are no longer passive components but autonomous actors with their own architectural opinions and the ability to act on them. In addition, traditional architectural patterns break when services become goal-seeking agents. Unexpected coupling emerges, contracts drift autonomously, data consistency becomes unpredictable, and governance blind spots multiply. Therefore, we need new patterns that let agents optimize while preserving system correctness.

This session tackles the fundamental challenge every architect will face in 2026: How do you maintain architectural integrity, ensure system reliability, and validate correctness when your services are autonomous actors that can redesign themselves?

In this session you will learn a taxonomy of agent-driven boundary changes with concrete examples, architectural patterns that preserve integrity while enabling autonomous evolution and testing methodologies specifically designed for non-deterministic and self-modifying systems.