Authors:
T. Pranavadit, N. Sankar Ram, M. A. Bala Kumar, R. V. Vishal, R. Vinoth, Mahmoud Abuzaid, Malak Ismail, Manar Messed, Mazen Mawla

Addresses:
Department of Computer Science and Engineering in AIML, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India. Department of Computer Science, Zewail City of Science and Technology, October Gardens, Giza Governorate, Egypt.

Abstract:

The paper proposes a prototype of a multimodal AI workspace, W.E.D.N.E.S.D.A.Y., a union of text, voice, gesture, camera vision, and capacitive touch that would be projected onto a tri-layer interactive display via a holographic mixed-reality interface. The system combines multi-LLM routing with cascaded fallback, YOLO-based real-time vision, and a three-path voice pipeline combining Whisper, Gemini Audio, and two TTS engines. UI, application services, AI intelligence, and GPU-based persistence will provide layers that enable real-time interaction and scale. Low-latency inference is experimentally proven with a maximum vision processing rate of 124 FPS and an average response time of 1.5s in a local-cloud deployment. The system presents an agent-coordination system that assists with autonomous web automation, file manipulation, and tool execution. Findings show strong fault tolerance, high throughput, and smooth multimodal fusion of interactive AI workspace applications in real-world computing environments, successfully applied in assistive and industrial situations and deployment environments.

Keywords: Holography Interface; Edge Computing; IoT Workspace; Agent Systems; Gemini Audio; File Manipulation; Deployment Environment; Real-World Computing.

Received on: 02/07/2025, Revised on: 27/08/2025, Accepted on: 20/11/2025, Published on: 10/08/2026

DOI: 10.69888/FTSCS.2026.000722

FMDB Transactions on Sustainable Computing Systems, 2026 Vol. 4 No. 3, Pages: 196-209

• Views : 20
• Downloads : 4

Download PDF