Across Malaysia, enterprises are under pressure to deploy AI and transform processes and operations. To do so, however, they must move away from legacy infrastructure and into more cost-efficient and scalable environments.
During the “Modernising Databases and Scaling in the AI Era” roundtable, organised by Jicara Media and hosted by EDB Postgres AI, senior IT leaders in Malaysia discussed common barriers to AI deployment, including how to stop spending on dead weight expenses and start investing in innovation enablers.
Leaving legacy behind
Across the board, legacy infrastructure emerged as a common anchor tying enterprises down. One organisation, for example, began modernising its infrastructure in 2021, yet some of its applications are still legacy systems.
“We have adopted a hybrid cloud approach. We’re using SQL and other databases across more than 100 applications. The organisation itself has been around for decades, so you can imagine that there is quite a lot of data. We also have subsidiaries and quite a number of departments, so centralising data is quite difficult,” one of its senior managers said.
The same is true for asset management firm UEM Edgenta, which previously had many legacy applications that had not been touched for decades.
“Getting insights from these applications is tough, that’s why we want to modernise. However, we don’t want to limit usage to just a single point solution. As our people access the database, we also want them to gain institutional knowledge from it. But right now, my immediate concern is ensuring the availability of the applications during high usage or peak seasons,” shared Chua Yong Howe, Chief Digital Officer, UEM Edgenta.
Chua said they rely heavily on a major enterprise database system, alongside a cloud-based data analytics platform. Beyond these, the company is also supplemented by a combination of traditional relational database systems.
AI pressure
Regardless of industry, there is growing pressure, either from management or from customers, to deploy AI and modernise operations. Tze Phei Tee, CIO of energy solutions provider Wasco Berhad, recalled having the AI conversation with their CEO.
“He wants me to explore agentic AI. He said, ‘Let’s start with HR and finance, some of the back-office functions, and see whether agentic AI can help.’ Right now, we are still trying to identify the avenues for pursuing agentic AI. If we do find a use case we think is right, we need to understand how ready we are in terms of people and process transformation, from a technical standpoint, as well as what kind of stack we need to move to in order to support agentic AI,” he said.
Tze also raised workers’ willingness to be upskilled as a potential roadblock for AI transformation.
“Not all employees aspire to be upskilled yet. They are comfortable working on the routines they have been doing for quite some time, but this is a common challenge that all organisations will need to face and plan for,” he said.
According to Seah Boon Chong, Head, Sales Engineering, APJ, EDB, not all challenges require an AI solution. Drawing on customer experiences, he said starting with small use cases often yields quick wins that can later be scaled.
“You really need to find out, is it really necessary to do AI? Maybe not. Maybe analytics is what you need. Maybe AI is the next stage,” Seah noted.
Reliable support
As enterprises rely more on open-source database systems such as Postgres, one challenge is ensuring reliable support. Minor issues such as bug fixes, as well as major incidents such as service outages, require timely intervention. EDB Postgres AI positions itself in this space by helping organisations reduce vendor lock-in while providing enterprise-grade support for Postgres deployments.
“Open-source Postgres is very broad. But when we talk about mission-critical applications, you cannot afford any downtime. That’s where we come in. We provide enterprise tooling and support around Postgres to ensure the security and availability of your applications,” said Ang Lee Yen, Asia Sales MD, EDB.
“Especially for organisations governed by regulators, if you discover an issue, we can deploy a patch within minutes so everyone receives the fix,” Seah noted.
Data sovereignty presents another layer of complexity for enterprises running both on-premises and cloud environments. In many cases, organisations must manage siloed data, vendor lock-in, and complex workload orchestration while remaining compliant with regulatory requirements.
“When we talk about sovereignty, the questions are: where do we extract the data, how do we protect it, and how do we keep it within your control and within your country’s data centres? That includes large language models. Many organisations leave them on the internet, but we have to think about how to place them closer to your data, your vectorised information, and the RAG systems you retrieve,” Seah explained.
EDB Postgres AI offers a sovereign data and AI platform that combines the security and control of bare metal with the agility of the cloud, deployed in enterprises’ environment of choice.
The platform enables enterprises to run transactional, analytical, and AI workloads while keeping highly regulated data on-premises to meet compliance requirements. At the same time, other workloads can run in the cloud to take advantage of its flexibility and scalability, with both environments managed through a consistent experience across deployments.
Seah said this approach helps organisations reduce spending on infrastructure they no longer need and redirect those resources toward higher-value initiatives.
“If you don’t reduce the cost of your existing infrastructure, which consumes considerable resources, you will have limited capacity to invest in new initiatives,” he said.
