Mark Chong, Deputy CEO of Thailand’s AIS, talks to Contributing Editor Annie Turner about the drivers of automation, fighting the IT stack and the importance of speed.
Mark Chong became Deputy CEO of Thailand’s mobile operator Advanced Info Service (AIS) in December 2022. Previously he worked at Singtel for more than two decades, most recently in the role of Global CTO for almost six years. Singtel holds a 24.99% stake in Intouch Holdings, AIS’ parent company. Singtel is regarded as one of the world’s most advanced telcos, thriving in one of the fastest growing and most dynamic regions. Experience and expertise is definitely not lost in translation between the two.
Like Singtel, AIS has deployed 5G Standalone which Chong states “drives more automation because without automation, it becomes humanly impossible to handle all the various tasks”. He adds, “There’s a big difference between 4G and 5G: with 5G the network becomes programmable and there are more network elements for engineers to handle. When something breaks, how do you deal with that?”
“We really need to automate all the manual, mundane, trivial stuff to enable engineers to focus on the more sophisticated pieces of engineering,” Chong says.
At Singtel, Chong’s team created zero-touch trouble tickets, sending tickets into the field and began automating recovery of circuits to reduce the back-end’s workload. “We did things like automating our collection of firewall policies and their configuration to reduce labour – things that normally took a few hours came down to a matter of minutes,” he says.
AIS is taking a similar approach with trouble tickets to shorten recovery times, but first-call resolution is a big priority too. To help achieve this, data analytics of various network metrics are fed directly to call centre agents which also speeds up call handling.
Along with being programmable, 5G has other new capabilities, like orchestration, network slicing and the ability to expose network commands to third-parties including customers. “Manufacturing clients, for example, now want some say about how the network performs,” he explains. “5G networks can operate on commands generated from the customer’s network – under certain circumstances and conditions, of course – but when things break, where’s the cause?”
Chong gives real, commercial examples of interoperable scenarios: Singtel deployed its 5G infrastructure network solutions at a car manufacturing facility in Singapore. The radio coverage within the factory drives manufacturing processes. The same solution, which runs on Singtel’s Paragon orchestration platform, is used in vehicle manufacture in Thailand to monitor the power current.
AIS is also implementing a discrete, private 5G network for an open mining company. The above-ground network will drive autonomous excavators via communication links. He suggests a similar approach could be deployed in ports, which have many containers and also rely on automatic guided vehicles. However, Chong stresses, “Each customer has its own requirements so we have to listen to them and see how we can adapt our solution to suit their needs”.
Rethinking the IT stack
Interweaving customers’ networks into their own 5G infrastructure means most telcos must fundamentally rethink their IT stacks. “If we don’t, it is impossible to support all these new requirements at the required speed,” Chong says.
“The OSS needs to be the right set of metrics and measurements; processes need to be reinvented, rethought,” he continues. A simple example is that previously, field workers were not dispatched to a customer’s premises if something went wrong whereas now “we have started to employ enhanced radio in a car factory and the private network in the mine…I need to provide field operational support for these installations. I must change the structure of my support teams for customers,” he comments.
“We must change what we measure around that radio in the field and be able to pipe [operational information] to the customer’s dashboard because now business mission-critical processes run on the network.”
How far is AIS on its network automation journey? He replies, “I’d say it’s in its early stages but moving in the right direction. One of the issues is our monolithic IT stack. To introduce new capabilities requires investment in resources and quite a bit of effort. It’s not just about having a modular OSS, but the whole modular IT stack, including billing and your front-end digital shop – wherever you’re seeing customers, from the cradle to the grave, in the BSS and for omnichannel.”
APIs and making radical changes
These radical moves cannot be made overnight. AIS’ approach is to introduce application program interfaces (APIs) in certain segments of the stack to link various parts for internal use. He elaborates, “Likewise, you need open APIs to interoperate with the customer’s network, to introduce new capabilities, new elements into your network.
“As a brownfield operator, we need a new reference architecture, so we’re running an architecture at two speeds. You have to create a new mediation layer or new enterprise service bus and progressively migrate elements onto it. The telcos must control the architecture, the enterprise service bus and the open, standard APIs to connect new boxes to the bus. You can get an outsourcer to build it, but the CIO or CTO must control which elements come into their network and how you hook them up.”
What does he mean when he refers to open APIs? Chong explains, “Singtel is a member of the GSMA board and one of the founders of its Open Gateway. Prior to that, we had embarked on the adoption of TM Forum’s Open APIs.
“One of the things we are very loud about is that CAMARA [which is part of the Open Gateway initiative to support abstracting from network APIs to service APIs] should be compatible with or should absorb TM Forum’s APIs into its larger set because many telcos and companies have already adopted them and we should not reinvent the wheel.
“We should focus very much on having a unified, harmonious, federated outcome that enables companies to go to market quickly and introduce new use cases. It’s less about who has the right technology, but rather, let’s bring the benefits of 5G’s programmability and operability across various systems and platforms to benefit customers.”
Becoming cloud native
This pragmatism extends to AIS’ approach to cloud. “We are largely working with our network vendors as we cannot do it all by ourselves – and they have gone into telco cloud. Are they ‘truly native’ like the OTT hyperscalers? I would say not to the same level…but good enough,” Chong says.
He points out that the various global cloud and network equipment providers each have their own approach. “Some of them tried to be the universal transporter or handler of workloads from one cloud to another. VMware does that quite well, but the rest are pretty proprietary,” Chong says.
“Each cloud provider brings the benefits of agility, resiliency, scale but in their own proprietary way and the commercials of each cloud are different, so you must go in with your eyes open.…You have to think through what could go into the cloud and what should not. If you have to refactor applications [for them to become cloud native] that’s a big chunk of work. Some are better left alone; for others with a longer runway, it’s worth the effort.”
AI to benefit customers
How much will the use of AI, including generative AI, be instrumental in automation and delivering benefits to customers? Chong says while he is very interested in generative AI’s potential it is immature but he believes it will raise productivity.
For now, machine learning (ML) is the most common type of AI used by AIS (and many other telcos), such as “in the go-to-market piece,” Chong says. “We analyse data to build as much of a 360-degree customer profile as possible, then we cook up campaigns and come up with bundles for next best offers” proffered via the digital channel or store the customer choses.
Chong thinks AI could speed up handling queries from customers about promotions or their bills. “We will train a model to understand our policies and handle customers’ requests,”. He adds, “We have an AI-enabled voice robot that can conduct a human-like conversation when reminding customers about payment. We are in the final trial stages and it should be launched soon, for consumers and enterprise customers.”
ML also correlates alarms to engineers who can more quickly identify and focus on the root causes among the “ocean of alarms”. Singtel deploys cognitive technology to gather data about customers’ use of fixed broadband deployments and data analysis for preventive maintenance such as rebooting modems before their memory fills up, easing congestion.
The benefits of hindsight and speed
What would Chong do differently if he were starting over with network automation? He muses, “The end goals remain the same but we could have done better by bringing more people on board within the organisation, getting teams installed faster, learning new technologies faster. When we started, we were a bit tentative in trials and tests.
“We’re now more certain that we are on the right path. We’ve put things in place that have greatly improved our processes, reduced development time, etc. We’re now bringing on more parts of the organisation as they can see what we’ve done and are onboard.”
Mark Chong will be on the panel discussing Operational transformation: Accelerating the journey to zero-touch automation at FUTURENET ASIA in Singapore 18 & 19 October, at 4pm on Day 1.