- Homegrown AI solutions transform finance, legal, and customer service ops
- AIX 2.0 launch planned, eyes SME market for its internally developed AI applications
For the first time publicly, CelcomDigi showcased a comprehensive AI ecosystem of internally developed solutions at the recent ASEAN AI Summit – ranging from solutions which automate hundreds of business processes, reduce multi-day tasks to minutes, and even one which handles the equivalent of 50 employees’ worth of work through a single digital assistant named Sophia.
And if this sends a chill down your spine, wondering about job security, CelcomDigi says there have been no retrenchments. “It’s always about enhancing productivity and upskilling individuals,” Kugan Thirunavakasaru, its chief innovation officer emphasized when asked about job cuts.
The telco’s demonstration of some of its AI applications reveal, not a patchwork of AI solutions pushed out by eager managers aiming to notch efficiency gains, but rather a transformation strategy that has been quietly unfolding.
CelcomDigi’s “fusion” strategy
In an interview with Digital News Asia, Kugan outlined how the company is using a dogfooding approach to develop AI solutions internally first, adopt and improve upon them based on feedback, with a plan to eventually offering the various solutions to Malaysia’s small and medium enterprises (SMEs), a segment which represents 96.1% of the country’s business ecosystem (based on Department of Statistics Malaysia 2024 data).
“Most of this development is done in-house by our citizens to improve and automate processes, and enhance business efficiency,” Kugan said during a walkthrough of the CelcomDigi booth.
“For SMEs, they want to go digital, but cost is a big concern. Imagine if you can have a digital assistant to do your HR policy, automate all your reporting, your business rules, your processes – all can be enhanced by using AI to make it more efficient,” he explained.
The internal approach runs parallel to CelcomDigi’s external-facing AI Experience Centre (AIX), which focuses on co-creation with around 50 partners across eight verticals. While AIX showcases collaborative innovation, these internal solutions are part of CelcomDigi’s broader “fusion” strategy – positioning AI and data at the heart of the merged organization across all business functions. Recall that both Celcom and Digi were separate and fiercely competing companies before both their parent companies, Axiata Bhd and Telenor decided in 2022 that a merger would better serve their purpose.
The company uses AI for network predictive maintenance, automation, and customer service optimization through unified agent dashboards that eliminate the need for customers to repeat information across multiple touchpoints. This comprehensive integration targets core business functions including legal documentation, financial reporting, and compliance processes.
Cost structure and open source approach
CelcomDigi has taken distinctly different approaches to funding its internal versus external AI initiatives. While the company committed US$59.16 million (RM250 million) as part of its merger mandate to develop a robust innovation ecosystem through AIX and partnerships with over 50 ecosystem collaborators, internal AI development relies primarily on open-source platforms and cloud services.
“The internal AI development mostly is done through open source and does not cost much. You can go to Azure and develop this, and that’s the beauty of open source – you can do whatever you want and scale accordingly,” Kugan explained.
This approach enables rapid development cycles measured in weeks rather than months or years. The legal team’s SAGE project (read below), for example, moved from concept to production deployment within weeks, with continuous refinement based on user feedback.
Legal team leads digital transformation
The legal department has emerged as an unexpected AI pioneer, developing the SAGE (Scope Assembly Generation Engine) project that automates contract scope preparation. What previously required two to three days now takes minutes, with 70%-90% automation accuracy.
To the obvious question of whether its AI roll out triggered fear, uncertainty and resistance, Kugan said it actually generated enthusiasm among employees. “We are getting very positive feedback internally, because the buzzword today is AI and we don’t want to be left behind.”
By way of example, he said, “If your colleagues can prepare a scope of work in minutes versus someone who takes two to three days, you would obviously want to learn and adapt the former’s way.”
The legal team’s AI tool performs deep research capabilities that would typically require three to four days of human work, with a comprehensive analysis done in two to five minutes. The solution has been in production for three months, with continuous refinement improving accuracy from an initial 40%-50% to current levels of 70%-90%.
Digital employee handles 300 processes
Even more impressive is CelcomDigi’s most ambitious AI project called Sophia, a digital employee that automates 300 processes across the organization. Developed over four years starting in 2021, Sophia has evolved from basic automation in 2021-2023 to incorporating generative AI and agentic AI capabilities over the past two years.
“Roughly we calculated, she does about 50 people’s worth of work,” explained an exec during the demonstration. Sophia handles regulatory compliance, including automated reporting to the Malaysian Communications and Multimedia Commission (MCMC), and processes law enforcement data requests that previously required five to six days. Sophie (spine chill alert) – does it within an hour.
The digital assistant also manages onboarding processes, asset allocation, financial reporting, and daily operational tasks. Employees can interact with Sophia through Microsoft Teams for immediate responses, while the underlying automation engine continues processing background tasks.
Healthcare AI advances
Beyond administrative automation, CelcomDigi showcased healthcare AI applications that convert 2D MRI and CT scans into 3D visualizations for surgical planning. It is actively being used by 65 surgeons across multiple Malaysian hospitals including Universiti Malaya Medical Centre (a public and private hospital), Hospital Kuala Lumpur (public hospital), and KPJ Healthcare (private hospital group). It also completed Southeast Asia’s first surgery using this technology in collaboration with Universiti Malaya Medical Centre (UMMC), with support from Singapore’s National University Health System providing knowledge transfer. “We are trying to make UMMC as the hub so that other hospitals will learn from it,” a CelcomDigi exec noted during the demonstration.
AIX 2.0 and international interest
Kugan also shared plans for AIX 2.0, a refreshed version of the company’s external innovation hub. The original AIX attracted over 4,400 visitors from more than 240 companies and 14 universities, with commercial pilots increasing from 13 to 25 active projects. Delegations from Germany, Norway, Japan, and Uzbekistan have also visited the facility, with Kugan noting this interest emerged organically rather than through any active promotion on CelcomDigi’s part. “Most of the solutions you have in AIX are through international collaboration,” he said, suggesting global partnerships are driving some innovation directions.
Looking ahead
While some solutions like the unified agent dashboard remain in trial phases as the company gathers performance data across merged operations, CelcomDigi’s dual-track approach of developing AI internally while simultaneously building external partnerships through AIX reflects the telco’s ambitious plan to position itself as both an AI user and AI services provider to the market.
With AIX 2.0 launching soon and its SME commercialization plans advancing, the company’s internal AI maturity may prove as valuable as its external innovation showcase.
Related Articles
Keyword(s) :