
HARVIS :
Augmented Answering Machine
#LLM #Hologram #Interactive Visuals
Role
Master of Design Student
Team
Individual Academic Work
Timeline
4 Weeks (February 2025 ~ March 2025)
As part of the Designing Emerging Technologies course at UC Berkeley Master of Design, we were tasked to choose an old existing object and augment it using emerging technologies. I've developed HARVIS a Holographic Box as an augmented version of a digital answering machine.
Concept Map
I chose an old Digital Answering Machine as my object to augment. Answering Machines are limited to only playing back pre-recorded messages which is where I saw an opportunity to develop.
An augmented Answering Machine could feature :
Real Time Bi-lateral Communication (opposed to recorded Uni-lateral Communication)
Visual Representation of the Answering Figure (opposed to Audio-only)

System Architecture
The System Architecture consists mainly of the TouchDesigner Ecosystem being mirrored onto a phone screen from a PC monitor, which acts as a hologram as it is reflected onto a transparent sheet in a Hologram Box. This utilizes a holographic illusion technique called "Pepper's Ghost."

Fabrication
To maximize the clarity of the reflection I used black foamboards and a thin PVC sheet to fabricate the hologram box. The thinner the reflective sheet (PVC) the less distortion there is in the hologram. The wooden holder on top allows the user to position their phone in the most effective way to reflect their screen on to the PVC sheet.
Prototype Test

Iterated Final Version

Outcome
The final demo video features my holographic AI Agent, HARVIS trained on the database of my personal information, answering questions that I ask in real time.
Reflections
Through this project, I have gained expertise of implementing emerging technologies in the areas of Holograms and AI Interactions.
This includes :
Text to Speech Integration within TouchDesigner using Eleven Labs & ChatGPT
Hologram visualization using cost-efficient methods
AI Agent Training based on personal information knowledge bases
Copyright©2024. All right reserved Sun-Q Kim
