“If you look at GenAI, last year there was a lot of hype around potential, but this year it’s all about “Can you help us make it real?””

So Rohit Maheshwari observed. Maheshwari, Head of Products & Strategy at Subex, spoke to 6GWorld recently. While the next generation of telecoms networks gets a huge amount of attention, Maheshwari and Subex are focussed on a different kind of evolution.

What can we learn from this about the possible direction of telecoms?

Making It Real

“We are a niche fraud management, revenue assurance, asset assurance player. We also do a whole lot of partner billing and enterprise billing,” Maheshwari explained.

“Within our context, we see a lot of pressure on a) managing costs and b) at the same time ensuring customer experience is intact. For us, a telco tech evolution simply implies there are newer products, newer services which now need to be protected.”

While Subex might be involved in assurance and fraud management, this is fundamentally why the company is stepping up to help “make AI real” for telcos – principally as a response to demand.

“We have been sitting on a massive amount of Telco data. CIOs, CFOs, CMOs have all been saying, “Hey, you already sit on very high resolution, very fast data. What can you do with this data?”” Maheshwari explained.

“So our long-term aim is to help unlock the value of rich data that Telcos sit on by leveraging AI where necessary and help them deliver connected experiences to their customers. It doesn’t matter whether it is 3G or 6G as the underlying technology. While there are implications, from an intent perspective it doesn’t matter to us.”

Step one of this AI shift, Maheshwari says, is “institutionalising very complex forensics and fraud investigation methods by taking it off the shoulders of individual experts, turning it into task-optimised Large Language Models and automating the whole process.”

This might be seen as a double-edged sword by fraud experts – while it might be seen as a benefit to democratise that expertise, it reduces the opportunities for human experts. However, as pointed out in this article, AI (including AIs being used within telcos) is liable to behave in unexpected ways including in fraudulent ones. Keeping pace with their capabilities and policing them is also going to require automation and AI.

That does, however, open up a parallel opportunity, as Maheshwari pointed out: “Unlocking very rich intelligence which can be monetised by Telcos externally.”

What does that mean in practice?

“In emerging markets telco behavioural data can be a good proxy for creditworthiness,” Maheshwari explained.

He is, of course, referring to people who don’t manage money through established banks – depending on the market, that can often be a majority of people. In these situations, the most rigorous behavioural footprint available can be through the mobile. Assessing elements such as ability to manage telecoms microloans, regularity and size of top-up and so on can be tracked, and have been tracked for several years now, to create creditworthiness assessments and enable loans where a ‘traditional’ credit rating cannot be assessed.

“For us, it’s all about unlocking that data as a proxy for creditworthiness and enabling unbanked customers to take new loans or financial services, but while doing that, carefully guarding and protecting what the customer doesn’t want to share,” Maheshwari commented.

Emerging markets have been a ripe marketplace for this kind of business, not least because there has been a shortage of established financial players in these markets.

However, there is no compelling reason why it should need to remain only a capability in those markets. The telecoms provider has, in many places, a unique perspective on the activity, location, preferences, network and behaviour of their clients. Concern about how data could be gathered and shared safely was one of the elements that scuttled the spread of Location-Based Services in the 2010s; however, we have come a long way since then.

“It’s about consent-driven sharing of data,” Maheshwari noted.

In an environment where telecoms providers realistically both want to train AIs on live data flowing through their networks and open up APIs with location data and more, it’s clear that the discussion has moved forward a good deal. Technology and regulation are starting to align better. The opportunities to leverage that are only starting to open up.

Filling The Trust Void

“With digital, the possibilities of new services and products are endless,” Maheshwari agreed. “Which also opens doors for a lot of customer confusion about what exactly they are subscribed to and what they are being charged for. And in many operators, the number one reason for reaching out to contact centres, or indeed for poor customer experience, is because of mistrust on what they’re being charged.”

This kind of problem is symptomatic of people’s experience an increasingly complex online environment – and also the challenges involved in rating and billing for a complex combination of different elements.

Maheshwari mentioned this in the context of providing clarity and assurance on the billing and customer case side, which is absolutely an important issue. However, it ties into something more fundamental in consumers’ online experience.

Last year, 6GWorld collaborated with Digital Mind State exploring different aspects of “Life Beyond 5G”, which brought together not only experts in telecoms, AI and robotics but areas including parenting, workplaces, music and relationships. While each of the sessions had different focuses, a prevalent theme was concern over how to navigate safely in a digital environment which is fundamentally alien to how we’ve evolved.

From the parenting guru training family members to spot deepfakes to the hip-hop artist assuring fans that a particular avatar truly represents them, there is a clear demand for transparency and the ability to build trust in an online environment.

Maheshwari concurred: “If you’re in a crowded room, you look left and right and for sure there is someone who has fallen victim to fraud. And I think there is enough data and statistics to suggest each one of us in our lifetime is going to fall victim to some fraud or other.”

Can that translate into a financial benefit for telcos? Subex, and the TMF more widely, certainly hope so. Maheshwari described how they have been working on a new open API which could start to do just that.

“It’s called the TMF 770. 770 is to share fraud intelligence based on telco data with other industries. So for instance, if I’m a bank and if I worry about a transaction,

I can interrogate API 770 find out if the phone number associated with the transaction has recently undergone a SIM swap, which would be a potential account takeover.”

That sounds helpful, though if that was all it was, that would be very limited. However, more data can be shared:

“What is the longevity of this connection? Is it a recent connection or has it been in existence for several years? What about the device? Is this a new device or an existing device? What about the location? Is it currently latched to home location or seems to have gone way beyond where normally I would go? Oh, they can also access a subjective assessment about the account status – is there a high risk of fraud so far as the Telco thinks, with their reasoning, or there is none and it seems to be all good?”

In all of this, notably, there isn’t ay identifying data going through. It’s not “Where is the person?” but “Are they where they’d be expected to be?” Not “What device is it?” but “Did it change recently?” All of this is about identifying potential behavioural indicators of fraud rather than personally identifying anyone.

This kind of insight could create highly valuable information for a wide variety of different contexts; not only for a bank transaction, but – in the case of our earlier hip-hop artist – also help assure other consumers of their avatar’s bona fides.

Is this a magic bullet? Probably not, but it’s a start. As Maheshwari concedes, “As creative as Telcos are in offering services, the people who want to game them are equally, if not more creative.”

Image by Mohamed Hassan from Pixabay