Design thinking was built when humans were the only actors. That's no longer true. The methodology needs to catch up.
The classic design loop — empathize, define, ideate, prototype, test — assumes one class of actor: the human user. That assumption just broke.
Autonomous agents aren't just tools — they're participants. They plan. They act. They fail in new ways. Designing for them means mapping not just human needs, but what agents should be empowered to do, where they must stop, and how humans and agents share a problem space.
AI engineers build the infrastructure. Design researchers study human-centered methods. No one sits at the intersection. This workshop series combines RedClaw's live agent infrastructure with SCD's rigorous design methodology to build a new discipline from the ground up, rooted at UIUC.
Hands-on workshops combining live infrastructure demonstrations with rigorous human-centered design methodology — for SCD scholars, graduate researchers, and practitioners at the frontier of autonomous systems.
The empathy phase gets reimagined. We map not just human needs but agent capabilities, failure modes, and contested spaces between. Saadeddine's collaborative problem-solving research maps directly onto human-agent task sharing.
RedClaw's live gateway as the canvas. Harness engineering defines the permissions, memory, and feedback loops that make agents safe and useful. Constraint design isn't just safety — it's a profound design decision.
Most systems bolt AI onto existing workflows. AI-native design asks: if agents were first-class actors from day one, what would this look like? Real RedClaw deployments across agriculture, medicine, and science.
As agents pursue goals autonomously, governance sharpens: who sets the objectives? The facilitator's role becomes more consequential, not less. SCD's facilitation methodology becomes the design lens for oversight.
MCP and Agent2Agent are now open standards under Linux Foundation. The plumbing for multi-agent coordination is standardized. Design thinking hasn't kept pace.
Self-hosted, local AI stacks are where institutions are heading for compliance and data sovereignty. RedClaw is already there — designing for these environments is a new discipline.
Most organizations are stuck in pilot mode not because the tech fails, but because the design frameworks don't exist. 40% of agentic projects are projected to fail by 2027.
The Illinois Quantum & Microelectronics Park, SCD, and RedClaw's work converge here. The infrastructure and methodology are already in the room. No competitor has this.
RedClaw builds advanced AI agent infrastructure for industries where data velocity meets mission criticality — agriculture, scientific research, and medicine. The RedClaw Gateway operates with sub-50ms routing latency across 10+ channels, fully self-hosted and air-gapped.
Saadeddine leads research at the Siebel Center for Design studying collaborative problem-solving and the role of facilitation in design-based STEM learning. His work on how students and teachers navigate shared problem spaces maps almost directly onto human-agent collaboration.
Four-module curriculum for SCD scholars and practitioners. Begins with a single 90-minute provocation session.
A publishable body of work establishing Agentic Design Thinking as a recognized field — papers, proceedings, case studies.
A publicly available framework other institutions can adopt — positioning UIUC as the origin point of this new discipline.