ABOUT US
GoldboxAI focuses on the development of Vertical AI solutions to solve industry-specific bottlenecks and business pain points for enabling forward-thinking enterprises to supercharge their business growth. Our areas of expertise range from data preprocessing and predictive analytics to computer vision and generative AI. We own proprietary technologies and preemptive experiences to engage into digital transformation with hundreds of IoT and unmanned projects being deployed regionally, now evolving into a new era of AI-powered solutions for intelligent enterprises.
We develop AI-powered solutions for enabling Intelligent Enterprises
AI in Retail​
-
Retail shelf analytics using mobile scanning
-
Self checkout loss prevention with edge intelligence
-
Image captioning for e-commerce product search
​ AI in Education
-
Creation of learning materials featuring Avatar instructors
-
Automated assignments grading and assessment
-
Personalized learning and virtual tutoring platform
​ AI in Finance
-
Financial chatbot with multimodal RAG
-
Anomaly detection by screening large datasets
-
Automated ETL (extract, transform, load) pipeline
​ AI in Healthcare
-
Health virtual assistant with multimodal RAG
-
Predictive analytics for early symptom detection
-
Movement analysis to assist physical therapy
​ AI in Government
-
Multimodal review of large volumes of documents
-
Automated processing of citizen application of public services
-
Prediction of public opinions on newly launched policies
​ AI in Transportation
-
Road safety monitoring with AI and GIS integration
-
Analysis of driver behavior inside vehicles
-
Prediction of traffic congestion for alternative routes
​ AI in Environment
-
Compute vision to separate components in recycling
-
Detection of plant growth through image analysis
-
Predictive analytics for fire incidents and public hazards
​ AI in Industrials
-
Analysis of unsafe scenarios for worker safety monitoring
-
Detection of non-conformities and defects in finished products
-
Predictive maintenance for early sign of machinery fatigue
​ AI in Sports
-
Player movement and performance analysis
-
Automated creation of sports highlight reels
-
Movement analysis for fitness training and personal workout
Vertical AI for Intelligent Enterprises
Empowering Business Transformation with AI-powered Solutions
We develop AI-powered tools and applications to help organisations analyse complex datasets, identify patterns, and generate insights to effectively solve business pain points and drive digital transformation.
OUR INNOVATION
AI is transforming the way we live and work. It can analyze vast amounts of data which human cannot handle with ease. It can also synthesize new knowledge, generate fresh ideas and uncover hidden patterns. In particular, the​ rapid development of LLMs and agent-based approach have significantly transformed the field of AI, demonstrating remarkable capabilities in multimodal functionality and decison intelligence. Our innovation efforts are directed towards various domains, namely predictive analytics, computer vision and generative AI powered by our adaptive models for multimodal data processing and retrieval augmented generation.
Data Preprocessing
Transforming raw data into more structured representations suitable for data mining and machine learning.
​
Involving the handling of missing values, removing duplicates and filtering outliers to derive meaningful variables and standardized features.
Data Annotation
Adding meaningful labels and tags to raw data such as images, video and audio for ML model training.
Typical approaches include bounding boxes, semantic segmentation, keypoint annotation, video object tracking and audio emotion detection.​​​
Predictive Analytics
Analyzing historical data with statistical analysis and machine learning to predict likelihood of future outcomes.
​​
Common techniques include multivariate regression, classification and clustering for supervised and unsupervised machine learning.
Computer Vision
A critical field of AI that enables machines to interpret and understand visual content that mimic the human eyes.
​​
Core techniques range from OCR image-to-text conversion to applying deep learning models for object detection and image recognition.
Generative AI
Generation of new content leveraging neural networks to identify patterns and structures within existing data.
​​​​
Applicable to typical use cases such as document writing and artwork creation using popular LLM models like ChatGPT, DeepSeek and alike.
Agentic AI
A new class of AI to empower agents to make decisions and take non-rule-based actions with reinforcement learning.
​​
Combining multiple techniques such as natural language processing, machine learning and computer vision, depending on the environment.
OUR PRODUCT
AI products are software or hardware solutions designed to automate processes, transform workflows and enhance decision-making by leveraging artificial intelligence techniques. They are not simply smart modules or boxes but rather intelligent tools with high degree of customization to fine-tune the AI for specific use cases. We continue to collaborate with university researchers and industry practitioners to focus R&D on distilled models for quicker adjustments to evolving customer needs and market conditions. This can provide valuable learning opportunities and insights for large-scale projects in the future, enabling us to build expertise in specific domains and develop better end products.
AI Product Analyzer (ApaBOX)
​Conventional approach for stocking and selling product involves assigning a unique barcode to each product, which is then printed onto a label. But manually handling massive volume of products is labor intensive. Based on edge intelligence, ApaBox makes use of image inference and label scanning to augment product recognition.
​​​
Designed for supporting shelf analysis and self-checkout for unmanned retail
AI Document Analyzer (AdaBOX)
​Multimodal content in text, image, video and audio formats are often stored for recording purposes. Once these content are aggregated into massive volumes, retrieval can become too challenging. AdaBox is based on OCR imaging and RAG-based LLM to extract and reconstruct useful content from multimodal input.
​Suitable for organization that store huge volumes of multi-format documents
AI Movement Analyzer (AmaBOX)
​Recognizing the human gestures or body movement is crucial to understand the needs and behaviours in different context such as physical therapy and sports training. AmaBox combines the algorithms from pose estimation and predictive analytics to detect and analyze human body movement.
Built for enabling automated analysis of body gesture and movement
Our proprietary technology patents
An automated system and method for retail self checkout with AI-powered product recognition based on hybrid cloud architecture
(Patent No. HK30088199)
Automated cleaning system with collaborative robot and AI vision for F&B kitchen automation
(Patent No. 10202251179V)
A Smart telehealth system monitored with context-aware sensing and multi-sensor data fusion for self-health monitoring and remote diagnosis
(Patent No. HK30049803)
System and method for multi-currency management using cash processing kiosks based on machine learning and distributed infrasturcure
(Patent No. CN107393212)
OUR SOLUTION
AI is evolving at an unprecedented pace with the launch of new foundational models pre-trained with massive datasets. These adaptive models such as GPT, Gemini, Llama and Deepseek, have transcended AI platforms by enabling multiple applications across diverse industries. However, most platforms present significant challenges including the need for extensive computational power, high energy consumption and complex software optimization in model training and inference. Our approach to AI-powered solution design is based on distillation and acceleration which adopts a hybrid and distributed machine learning and inference architecture. This enables the removal of redundant computations and to train model without overfitting so as to achieve higher efficiency and better resources utlization.
Here are some featured projects:
Bank customer onboarding with eKYC and face matching
HR resume translation and enhancement using GPT-4 LLM
Education student learning evaluation using GPT-4 Turbo
Facility alarm for fire and smoke detection using YOLOv8n-seg
Bank document privacy management using Llama 3-70B LLM
HR CV semantic search using OpenAI text-embedding-3-large
Healthcare chatbot based on RAG using Mistral-7B LLM
Factory material verification using OCR AI image-to-text
Bank customer data integration for ML predictive analysis
HR interview analysis using Whisper AI and GPT-4 LLM
Digital image auto captioning using Vision Transformer
Architecture wood pattern recognition using ML image analysis
Book a free consultation with us to explore the potential of your businesses with AI
Contact Us
Unit 1905, 19/F, Ho Lik Centre, 66A Sha Tsui Road, Tsuen Wan, Hong Kong
105 Cecil Street, #20-1 The Octagon,
Singapore 069534
Block B, B-1, Sunway PJ51A, Jalan SS9A/A9, Seri Sertia, 47300 Petaling Jaya, Selangor, Malaysia
​
17th Floor, MD Complex Office Building, 68 Nguyen Co Thach Street, Hanoi, Vietnam
