Agentic AI Design: Beyond Screens, Into Systems

Salesforce’s 2025 review reveals a profound redefinition of design’s role, shifting its focus from traditional interfaces to architecting the very intelligence of AI. This evolution, particularly in agentic AI design, positions designers as central to building trustworthy and effective AI systems. The core challenge now lies in how AI understands the world and human interaction, not merely how it appears on a screen. This strategic pivot underscores design’s critical function in transforming AI potential into tangible business value.The era of agentic AI design demands a fundamental re-evaluation of user experience principles. Traditional UX frameworks, built for static interfaces and linear click paths, prove inadequate for dynamic, multi-agent orchestration. According to the announcement, Salesforce’s Chief Experience Officer, Kat Holmes, emphasizes treating AI agents as a “new kind of user,” prioritizing human goals and requiring flexible, adaptable systems. Designers are now tasked with crafting experiences that harness coordinated AI agents across platforms while preserving human agency and comprehension. This necessitates an “invisible scaffolding” that enables real-time AI responses, surfacing relevant information precisely when and where it is needed. This paradigm shift mandates a new skillset for designers, moving beyond visual aesthetics to deeper architectural considerations. Kat Holmes implores CEOs to invest in critical design skills such as experience architecture, agent collaboration, conversation design, and building context maps and ontologies. The industry has already seen “vibe coding” blur the lines between designers and developers, fostering rapid prototyping and shared capabilities. Crucially, robust agentic design systems, exemplified by Salesforce’s SLDS 2, provide the decoupled architecture and modular components needed to build and scale these dynamic experiences, underscoring the imperative of systems thinking over isolated screen design. Architecting Meaning and Human CollaborationEffective agentic AI design hinges on a deep understanding of human communication patterns and expectations. Conversation design becomes paramount, shaping AI behavior to align with human interaction styles and providing strategies to repair interactions when they inevitably falter. Research indicates that AI agents must prove genuinely useful, not just functional, integrating seamlessly into human workflows to improve productivity. This extends to the design of “multiplayer” experiences, enabling more fluid collaboration between humans and AI agents, moving beyond tools that primarily support independent work. Crucially, mapping meaning and context through ontology is foundational for AI to understand the language and nuances of a business. Without this “system for organizing the data about the data,” AI agents would operate in chaos, unable to make informed decisions based on accessible information. Designers must become “semantic thinkers,” asking not just how a button looks, but what the system understands about its purpose, context, and behavior. This makes context engineering the new design material, ensuring that AI agents are grounded in relevant business knowledge rather than shipping “broken interfaces in a clever new wrapper.” The implications for the tech industry are significant: design is no longer a downstream aesthetic concern but a strategic imperative for AI’s success and adoption. By embracing a design-led approach to agentic AI, organizations can transform AI’s raw potential into practical, trustworthy business value. This ensures AI understands human intent and goals, responds predictably, and ultimately makes sense in complex operational environments, solidifying design’s role as the crucial bridge between AI capability and human usability.