Agentic Systems Design™
Future work will be published on www.agenticsystemsdesign.com
Traditional systems design assumes a world of predictable components and human operators.
But today's intelligent systems are made up of multiple forms of agency — human, artificial, and institutional — each capable of making decisions, learning, and adapting on its own...
... A customer service AI that learns to escalate certain issues, a recommendation system that shapes user behavior, a team member relying on AI-generated insights — each represents a form of agency that changes how systems actually work.
This reality changes how we design and build systems.
We are no longer designing tools that wait for input, but systems that think, act, and collaborate alongside us.
The challenge is not just technical — it is organizational, strategic, ethical and even philosophical. How do we design interactions, workflows, and structures that keep autonomous systems aligned with human goals?
Agentic Systems Design™ is the framework I'm developing to address that question.
It combines systems thinking, AI architecture, organizational behavior and design strategy to help builders — product leaders, system architects, researchers — create coherent systems where human and machine agency reinforce rather than replace one another.
In practice, it means designing for coordination over control; transparency over opacity; adaptation over enforcement. At every layer — product, organization, or ecosystem — we ask: what agency are we introducing, and how will it interact with the rest?
This will not be simply theoretical. An Agentic Systems Design™ Framework should give builders what they need today — practical guidance for shaping systems where humans and intelligent agents work together with clarity, purpose, and alignment.
If you're building, designing, or researching in this space — and asking similar questions about how humans and intelligent systems can work better together — I'd love to connect. I'm looking for collaborators to develop case studies, refine the framework, and build a community of practice around these questions.
Email me at: ASD (at) melvinyuan (dot) com
Future work will be published on www.agenticsystemsdesign.com