skip to content
copy link
1 min
exploring voice-driven interfaces grounded in real data
project type
evaluative, ux-focused research
company
tecmie
industry
artificial intelligence
role
user experience research
date
—
present
at a glance
background
imagine a drag-and-drop website builder; it employs pre-built, pre-coded components that you can piece together, and you get what we can call a custom website...
with the advent of chatgpt, claude and other ai tools, one can easily request (prompt) a fully built website, and even with the same prompt repeated a second time, you get different structure, copy and design.
now take this a step further: imagine a financial app whose core is ai. this application generates user interfaces on the fly with user input (voice). you could say, "show me a chart of my spending and generate actionable insights from my spending habits, but make it look fun and easy to understand." it would create a dynamic, colorful, functional and accurate graph for you, providing factual, not hallucinated insights.
that was the purpose of this research; to theorize how generative user interfaces and user experiences will function with voice and real, non-hallucinated data. the backbone for this experiment is the model context protocol, similar to the mvc (model, view, controller) and mvvm (model, view and view-model) models of building on the web, but with ai at the core.
