Human-Agent-Experience (HAX) Design
Designing for the internet of agents
A framework for guiding how people and autonomous systems work together
The human-agent experience
We’re moving beyond assistants and copilots. Today’s agents act with more autonomy, make decisions across systems, and collaborate with humans in more complex ways. Designing for this shift means rethinking how people and AI interact.
At Outshift, we’ve developed a set of principles to help teams design agentic systems that are trustworthy, collaborative, and clear in how they work.
A Research-driven framework for agentic design
These five principles emerged from studying how people actually interact with agentic systems—what works, where trust breaks down, and how collaboration succeeds or fails in practice.
You can dive deeper into the patterns, failures, and design insights in the full HAX white paper.
Guiding principles
Control
Humans guide interactions. Agents adapt to human intent and follow their lead through every step.
Clarity
Keep partnership transparent. AI communicates actions clearly, using progressive disclosure to reveal more detail when needed.
Recovery
Own the mistakes. When Agent gets it wrong, it helps users identify the issue, offer clear paths to correct it, and learns from the experience.
Collaboration
Work together towards shared goals. Agents contribute actively while staying responsive to user preferences and feedback.
Traceability
Build understanding and trust. Agent makes their reasoning visible so users can see how decisions were reached.