/ After 15 years inside Malaysia's largest enterprises, the team behind Glem.ai is turning AI ambition into measurable results
It started as a data solutions provider. Today it is one of the region's agentic workflow specialists, taking large institutions from the first conversation to results they can measure.
KUALA LUMPUR, 17 November 2026 - Most large organisations no longer need convincing that AI matters. Their problem is what comes next. Pilots stall before they scale. Departments pick up their own tools. Processes that took hours still take hours, and leadership struggles to see what any of it has delivered.
The team behind Glem.ai has spent about 15 years solving problems like these from the inside. Over that time, it has consulted for and delivered solutions to some of Malaysia's largest companies, first as a data solutions provider, then increasingly in AI. That experience has shaped its view that technology alone rarely changes how an organisation works. The process has to change too.
"Companies don't come to us looking for another platform. They come because a process is costing them too much time, too much money or too much risk," said Andy Leo, Director of Global Sales at Glem.ai. "We've spent years inside these organisations. We know where the work actually gets stuck. Our job is to fix that and prove it in the numbers, not hand over software and walk away."
The company's journey mirrors the one its clients are now on. Years of building data foundations for large enterprises gave the team a deep understanding of how information moves through an organisation. As AI matured, that understanding became the basis for designing agentic workflows, where AI agents take on multi-step work across systems rather than answering one question at a time.
Along the way, the team built its own technology, Glem.ai, to run and govern those workflows. But Glem.ai is the means, not the offer. What clients buy is the outcome.
Glem.ai approaches agentic AI through business process re-engineering. Rather than selling software and leaving clients to work out the rest, its forward-deployed engineers work alongside client teams to rethink how work gets done, then use operational results to guide further improvements:
1. Understand the work. Map the current process, identify bottlenecks and establish a baseline for time, cost and quality.
2. Redesign the process. Remove unnecessary steps and define how people, agents and systems should work together, including where human judgement and approval are needed.
3. Build the foundation and workflows. Prepare the data, develop the agents and connect the redesigned process to existing enterprise systems.
4. Deploy and govern. Put the workflows into operation inside the client's own infrastructure, with appropriate access controls, human oversight and auditability.
5. Continuously improve. Measure performance against the baseline, learn from exceptions and user feedback, and refine the process, data and agents. Apply those lessons to further use cases.
"Large institutions don't need another vendor. They need a partner who will own the result with them," Leo said. "That's why we measure every engagement against a baseline, and why go-live is never the finish line. Every exception and every piece of user feedback shows us how to make the process better and the next use case faster."
The team's client list reads like a cross-section of Malaysia's economy. It includes a conglomerate owned by the country's richest individual, Malaysia's largest highway concessionaire and Southeast Asia's largest aluminium manufacturer, alongside leading organisations in oil and gas, manufacturing, healthcare and government. Across these engagements, the impact reaches beyond individual tasks to how entire functions operate:
- Claims operations. Reviewing a single claim once took around three hours of manual work. With the process redesigned around AI agents, it now takes about six minutes, freeing specialists to focus on complex cases and exceptions.
- Sustainability reporting. Automating the extraction of data from utility bills cut a six-month reporting cycle to one, giving leadership a far timelier view of environmental performance as disclosure requirements tighten.
- Industrial operations. Equipment failures are now flagged before breakdown with 90% accuracy, shifting maintenance from reactive repair to planned intervention and protecting production uptime.
Taken together, the team's enterprise projects have cut processing times by as much as 90% and delivered hundreds of millions of ringgit in documented cost savings to date.
One successful workflow usually creates demand for the next. Because every engagement runs on Glem.ai, clients can see which AI systems and agents are in use, the data they access and what they cost, with access controls and full audit records throughout. Their data stays inside their own infrastructure, so the next use case starts on a foundation they already trust.
Glem.ai has a presence and active projects across Malaysia, the Philippines and the Middle East, with Beijing now providing a base in China.