ZETIC offers a platform that automatically optimizes and deploys AI models for on-device execution across any hardware, framework, or device. Their service streamlines the workflow from model upload through benchmarking to integration with a three‑line code snippet, reducing deployment time from months to hours. By leveraging CPU, GPU, and NPU acceleration, ZETIC enables low‑latency, privacy‑preserving AI without requiring model retraining.
Funding
Funding not disclosed
Founders
Product
Problem
Many AI applications rely on cloud-based GPU servers for inference, leading to high operational costs, increased latency, and potential security vulnerabilities. Transferring data to and from the cloud can also create bandwidth bottlenecks and privacy concerns.
Solution
ZETIC.ai offers a software platform, ZETIC.MLange, that enables AI models to run directly on edge devices powered by NPUs (Neural Processing Units), eliminating the need for cloud servers. Their automated pipeline transforms existing AI models into optimized versions for on-device deployment, achieving significant runtime performance improvements. By processing data locally, ZETIC.ai reduces operational expenses, minimizes latency, and enhances data security. The platform supports various operating systems and NPUs, providing universal compatibility for diverse hardware configurations.
Target Audience
ZETIC.ai targets AI service providers and companies seeking to reduce cloud server costs, improve AI performance, and enhance data security by deploying AI models on edge devices.
Features
- Automated model conversion pipeline for rapid on-device AI deployment
- Optimization for NPUs, delivering up to 60x faster runtime performance compared to CPUs
- Server-less AI architecture, reducing operational costs by up to 99%
- Support for various operating systems and NPU hardware
- Real-time face landmark, face detection, and face emotion recognition demos
- Object detection capabilities