About
Dominik Witzke is a Lead UX Designer at Star, a global design and innovation consultancy, and a lecturer at Hochschule für Gestaltung in Schwäbisch Gmünd. Before that, he shipped digital products at AutoScout24 and IXDS (now PwC). His work sits at the intersection of human-centered design, design leadership, and emerging technology — with a particular focus on what AI feels like to use.
Beyond Chat is the working surface for that focus. It started as a question — what should an AI interface look like when the assistant is not the metaphor? — and became a library of patterns and live prototypes that move AI work out of the conversation box and into operable interfaces.
The frame is the DOPE lens: Directability, Observability, Predictability, Explainability. Every pattern on this site is evaluated against those four axes, because the question is not whether AI can do the work, but whether the human can still steer, see, predict, and understand what it does.
Why this site
Almost every AI product converged on the same interaction model: a text box, a transcript, a stream. That model is fine for some tasks. For most of the work people actually do — editing documents, supervising agents, exploring options, verifying claims — it is the wrong substrate.
Beyond Chat is a working library of 80+ design patterns and live prototypes that show what those tasks look like when AI is moved out of the conversation box and into operable interfaces. Each pattern is paired with a chat failure mode, a working demo, and an evaluation against the DOPE lens.
DOPE Lens
Directability, Observability, Predictability, Explainability. Every pattern on this site is scored against those four axes — not "is the AI smart enough", but "can the human still steer, see, predict, and understand what the AI is doing".
Read the framework →In Public
The library grows alongside an ongoing series on LinkedIn — visual essays that pull apart specific products, principles, and failure modes in AI design. New patterns, essays, and prototypes ship roughly weekly.
Follow on LinkedIn ↗Get in touch
Consulting, lectures, conference talks, contributing a pattern, or just trading notes — I'm easy to reach. Email is the fastest path; LinkedIn works too.