AI interfaces may need less screen and more restraint, according to Jeff Easter, a human-computer interaction designer at PLAUD. Speaking on the Design Mindset podcast hosted by Radhika Seth on Yanko Design, Easter argued that the next wave of AI tools could be controlled by something as simple as a button, rather than by ever more crowded displays.
Easter's career spans data visualization, physical prototyping, and audio and hardware work, including his earlier contributions to the Amazon Echo Frames. That background shapes his view that the most useful AI products often succeed by doing less, not more. In his account, the design challenge is deciding when a device should stay out of the way.
One of the clearest examples he described involved notifications. In an experiment, filtering alerts so that only messages from selected senders came through reduced the urge to keep checking the phone. The result pointed to a broader principle: attention is easier to protect when a system holds back rather than demands a response.
The same logic guided a push-to-talk button project, which used a single physical control to start voice transcription. Easter said physical controls should give clear feedback about their state without relying on a display, so a user can tell at a glance whether the device is recording or idle. That kind of legibility, he suggested, matters more as devices shrink and screens disappear.
PLAUD's highlight button follows the same idea. It lets a user mark moments in a recording that are worth revisiting later. Those highlights then guide the AI toward what the user cares about most, while leaving the final judgment with the person. Easter emphasized that the human should remain the decision-maker, with the software acting as an assistant rather than an authority.
Easter also said AI outputs should fit into existing workflows instead of adding new management burdens. A tool that creates extra steps, files, or decisions for the user, he argued, undermines the convenience it is meant to provide. The goal is to make AI feel like a natural part of work already underway.
Where data is sent or committed, Easter said confirmation should happen on a screen. A physical button can start an action, but a visible review step gives the user a chance to check what is about to leave the device. Screens, in his view, remain useful for reviewing transcripts and source markers, even in a design philosophy that otherwise favors restraint.
Recording devices raise social questions as well. Easter noted that a recording device affects the people around it, and that visible indicators matter for bystanders. He drew a distinction between unobtrusive recording and concealed recording, arguing that a device can be small and quiet while still being openly signaled to the people in the room.
To illustrate where the approach could lead, Easter discussed a hypothetical concept he called a "totem." In this exploratory idea, placing the device face down would turn it off, flipping it over would start recording, and flipping it back would stop. The concept is not an announced PLAUD product, and Easter presented it as a way of thinking through physical interaction rather than a product roadmap.
The conversation left several questions open. The discussion did not address how well this approach works in practice, nor did it set out consent and permission norms for recording other people. Those questions will likely become more pressing as AI recording devices move from niche gadgets toward everyday tools, and as designers weigh how much control to hand to the user versus the machine.