AI could help emerging market operators improve field workforce productivity and build new enterprise services, but simply adding it to existing operations will not reduce costs, says Markus Persson, Global Telecommunications Industry Director at IFS.

Persson (pictured, below), a telecoms veteran whose career includes experience with Ericsson, said AI could help emerging-market operators improve field workforce productivity and develop new enterprise services. But simply adding AI to existing operations will not reduce costs. “AI by itself would not save them any money,” said Persson, IFS’s Global Telecommunications Industry Director.

The savings depend on what operators change as a result. That could mean retiring software licences, reducing reliance on subcontractors, or using existing staff to handle more work before expanding the organisation. Persson described downsizing the operator’s own workforce as a last resort.

“If you are to save money, you have to cut expenses somewhere,” he said. “If you just take AI and put it in and work like you always have, you just become a little bit more efficient. You will be able to do more, but it will cost you more.”

For operators in markets where margins are tight and extending coverage is expensive, the distinction matters. Persson said AI is currently more of a cost-cutting opportunity, although he expects it to become more of a revenue-generating one over time.

That shift could come through enterprise connectivity. Persson said operators could use IFS applications to monitor service level agreements for connectivity sold to factories and other businesses. He argued that operators have yet to realise the full promise of 5G for business-to-business services, in part because they may struggle to connect with prospective customers in other industries or speak their language.

IFS works with customers in manufacturing, mining, construction and engineering, among other sectors. Persson said that experience could help operators identify and serve enterprise customers. AI could also free up staff to pursue new business, he said, since developing an offer and winning initial customers may take a dedicated team months or even years. Once a service is established and repeatable, expansion should become easier.

Persson pointed to Singtel as an example of an operator that has grown its business-to-business sales and customer base across the region. He said few operators had achieved that kind of scale with 5G, while noting Singtel’s ownership of operators in other markets, including Optus in Australia and Bharti in India.

The opportunity may be particularly relevant in emerging markets, where mobile services play a larger role in daily life and the economy, Persson said. He noted that mobile money is more widely used in parts of Latin America, Africa and Southeast Asia than in Europe and North America, and cited MTN’s mobile money business as an example. Operators can be among the most reliable companies in some markets, he added.

Persson argued that operators in these markets could be quicker than their counterparts in Europe and North America to turn AI into revenue because they are accustomed to working with fewer resources. He compared the adoption of mobile money in African markets with its later arrival in Sweden.

Productivity depends on the workday

Field operations offer another potential source of efficiency, but Persson cautioned that software cannot create savings simply by reducing the time needed for individual jobs.

He said IFS had seen field workforce efficiency improve by 70 to 75 per cent when workers followed the jobs assigned to them. He also referred to improvements of more than 80 per cent in the United States. These figures describe efficiency gains, however, not necessarily equivalent reductions in costs.

If a technician saves five minutes on each job but uses that time for a longer break, the operator has not completed more work or reduced its spending. Savings emerge when the time saved is enough to fit an additional job into the working day. Persson described the effect as a “hockey stick”: productivity may rise without a corresponding financial gain until the team can complete another job.

The result also depends on conditions outside the software, including whether workers travel by car or bicycle and the state of local roads, he said. Operational gains shown in a presentation still need to be tested against practical conditions.

On infrastructure, Persson said operators moving to newer network equipment may encounter AI whether or not they choose to deploy it themselves. Equipment from vendors including Nokia, Huawei and Ericsson already incorporates AI for tasks such as self-configuration and adjusting parameters to suit network conditions.

He distinguished that embedded network functionality from AI used to change how operations and maintenance teams work. It is in those working practices, he said, that operators have more choice and where decisions about staffing, processes and spending will determine whether AI produces savings.

Security requires controls and human oversight

Persson said IFS had restricted the capabilities of its AI products in favour of security. He described earlier versions of IFS’s Intelligence product as having too much dependence on OpenAI, making it harder for IFS to control where data came from, where it went and how it was stored or used.

The company built additional controls around its products and data, he said. Intelligence does not store information unless a user instructs it to update something in IFS Cloud. The product also follows the user’s login permissions, so it can access only the data that person could access through the application.

Persson said IFS had held back some functionality to avoid breaching security requirements. He also said the company was working with its AI partners to determine how their respective tools could go to market in a way that provides value to the customer and gives each partner a fair return.

For operators considering where to introduce AI, Persson said he would be more cautious about using it in continuous integration and continuous development pipelines. Traditional workflows follow a set sequence, while goal-oriented agents can take different paths to reach an outcome. Some paths could be more damaging than others, he said. IFS is routing its agents through the same security controls used by Intelligence.

Persson’s central safeguard is that AI should not take consequential action without human approval. “As soon as it changes something, it never changes without the one who initiated it approving,” he said.

He rejected the idea that responsibility can be passed to an AI agent. “AI doesn’t start by itself. It’s always initiated by someone,” he said. The person who starts an agent remains responsible for its actions, in the same way a supervisor is accountable for an employee’s work.

Technical safeguards are only part of the picture, Persson added. Operators also need governance and a culture that encourages responsible use. “You can implement all the security you want in software, but if you have the door open, it doesn’t matter,” he said.