by Tan Aik Keong (AK)
As digitalisation keeps accelerating globally, artificial intelligence has stopped being just a tech-industry buzzword — it's become a core driver of business transformation and competitiveness. Malaysia is standing at a genuine inflection point in AI adoption right now. As an AI solutions provider, we see three categories carrying the most weight in the market today: generative AI, data-driven AI, and computer vision. Here's how each is actually showing up in enterprise transformation.
1. Generative AI: from language models to enterprise assistants
Generative AI, particularly large language model (LLM) applications, has fundamentally changed how humans interact with machines. From OpenAI's ChatGPT to Google Gemini to enterprise-grade models like Claude and LLaMA, generative AI can understand natural language, write content, summarise documents and even write code — strong enough at language understanding and generation to power customer-service automation, legal text analysis, marketing copy generation and more.
In enterprise deployments, we're seeing more local banks and telcos deploy generative AI as an internal knowledge assistant — pulling together SOPs, process manuals and regulatory documentation, and letting staff query it in natural language. That cuts information-lookup time dramatically and speeds up decision-making.
To make LLMs more useful in enterprise settings, retrieval-augmented generation (RAG) has become the dominant architecture. RAG combines embedding-based semantic search with a generative model, pulling relevant information from a knowledge base before generating a response, so answers stay accurate and contextually grounded. For any use case that needs to connect to internal systems, document repositories or product knowledge bases, RAG is the key technical path to building a genuinely useful enterprise AI assistant. We've also helped a manufacturing client build an AI-generated work-order system that automates machine repair requests and cuts labour costs.
The core technical building blocks here: fine-tuning, context window management, embedding-based retrieval, RAG architecture, and safety mechanisms like RLHF and content moderation.
2. Data intelligence: letting data drive prediction and decisions
Most established businesses sit on large volumes of structured and unstructured data whose value is routinely underused. Data-driven AI applies machine learning and statistical modelling to learn patterns from historical data and generate predictions and recommendations. In retail, for instance, AI can forecast next quarter's bestsellers from purchase history, location and market trends; in finance, data models can flag potential fraud and churn risk.
We built a forecasting engine for a pharmacy chain that combined sales records, seasonal holiday patterns and weather data, cutting stockouts by roughly 28%. In logistics, our dynamic route-optimisation system has saved clients over 15% on transport costs.
What matters most here is data-pipeline engineering, feature engineering, and building in explainability. A good data-AI project isn't just about the algorithm — it depends just as much on data quality and how tightly it's tied to the actual business problem.
3. Computer vision: seeing the data hidden in a scene
Computer vision gives AI the ability to "see," from static images to real-time video analysis, and it's now widely used in manufacturing quality inspection, smart security, retail analytics and smart-city applications.
In manufacturing, we've helped clients deploy deep-learning-based visual inspection systems that automatically flag product defects, bad solder joints or packaging anomalies. Using edge computing and high-frame-rate cameras, the system processes production-line images in real time and feeds anomaly data straight back into the MES system for closed-loop quality control.
Another example: mall foot-traffic analysis systems, where computer vision tracks customer movement patterns, dwell time and estimated age/gender, helping property owners optimise rent and store layout.
The core techniques here span object detection, pose estimation, segmentation, and deployment optimisation via frameworks like ONNX and TensorRT.
AI transformation isn't the future anymore — it's now
AI has moved out of the lab and into real business operations, becoming a key engine of digital transformation. Generative AI makes knowledge more accessible; data intelligence makes decisions more precise; computer vision turns the physical world into data. The real challenge for a business isn't whether to adopt AI — it's how to integrate it strategically for a durable competitive edge.
Malaysian companies should treat this as AI's "golden five years" — starting with small, focused pilots and scaling deployment gradually, while building internal AI literacy and technical capability. That's the only way to genuinely get ahead of this wave.
Part of the AK AI Corner column. Originally published in Oriental Daily (东方日报) on May 18, 2025.