Agents, Architecture, & Amnesia: Becoming AI-Native Without Losing Our Minds

AI in the software lifecycle is not a single leap — it’s a continuum of autonomy. We’ve moved from AI as a tool, to task-oriented agents, and now to multi-agent platforms. On the horizon are fully autonomous systems that promise speed and scale we’ve never seen before.

But here’s the risk: in the rush to adopt, too many teams are suffering from architectural amnesia. Governance fades. Trade-offs aren’t measured. Technical debt compounds at machine speed.

This keynote explores the AI Autonomy Continuum and what it means for architects, engineers, and technology leaders. At each stage — from copilots to multi-agent ecosystems — I’ll highlight the architectural principles that must not be forgotten: governance, experimentation, trade-off analysis, risk management, and resilience.

The message is clear: as systems gain autonomy, the role of architecture becomes more critical. Without disciplined guidance, autonomy amplifies risk just as quickly as it accelerates capability. Architects must design the guardrails that make each step toward autonomy safe, secure, and sustainable.