In 2024, 90% of telecommunications companies said they were using or exploring AI in some capacity. But the transformative, generative AI application that many people imagined would be taking care of entire telecoms functions is still a work in progress for most of the industry.
In reality, what we’re seeing today is actually trusted automation, analysis and clever algorithms leading to improved business decisions.
Origin-based routing (OBR) is a good example and is a hot topic to master for international voice operators and wholesalers. AI is not taking over this process, but it’s helping to significantly refine and optimize the operations that underpin it. This is what we at Telarix call origin-based handling (OBH) or origin-based optimization. This encompasses not just the task of routing, but the surrounding infrastructure of costing and pricing management, rating, billing, analysis, fraud management, and more.
So, what does the AI of today actually look like? And what do we want it to be responsible for? In this blog, we explore what AI is really equipped to handle, the best OBH use cases to apply it to today, and how this is setting the stage for more developments to come in the next few years.
AI and OBH – what’s hype and what’s reality
The major complexity in OBH is rate management and price list processing. Even within simple EU vs non-EU classifications, there could be hundreds of different rates to assign and manage. For a human, this is complex, time-consuming, and comes with a high risk of error.
This makes it a perfect use case for today’s AI capabilities. While AI tools aren’t yet ready to automate this entire function from end to end, decision-makers wouldn’t want them to. When a single mistake could cost millions, taking a human-in-the-loop approach is both less risky and more accessible.
That’s why the biggest successes we’re seeing are with functional AI tools like automation, algorithms, and intelligent analysis. Tools that are used to make the process more efficient – in terms of time, margins, and accuracy – but keep decision-makers in the driver's seat.
Here are the four main ways these kinds of tools are used in OBH today:
Fraud management
Although OBH helped telcos to increase margins, it also introduced new fraud incidents, such as number spoofing (or caller ID spoofing).
AI is adept at managing this new threat. Predictive analytics tools use historical data to forecast and identify fraud cases, which in turn supports routing decisions. This use case is one of the more mature AI applications in OBH thanks to its previous application, and therefore training, in other similar fraud cases, such as SIM box fraud, flash calls and Wangiri fraud.
Nevertheless, AI is adept at identifying previously unseen fraud types. By monitoring anomalies in OBH traffic, these models can detect and prevent fraud cases – even when no previous rule exists – before they cause widespread damage.
However, it will still take time and specialist training to introduce full-blown AI and automation into this domain. For example, to eliminate the currently high risk of false positives, which can be damaging to the telco’s bottom line.
Invoicing management
Invoicing involving OBH is an area where generative AI is starting to pick up momentum.
There’s plenty of automation already being applied to create new customer invoices based on historical templates. Usually, the content of invoices is highly repeatable, so it’s easy to automate this process.
However, the introduction of OBR in suppliers’ invoices makes the ingestion of new templates more complex – especially as templates and data can change without advanced notice,
However, machine learning (ML) and natural language processing (NLP) make it possible to automate these less repeatable cases as well. For example, predictive analysis can be used to analyze historical changes, learn patterns, and adapt templates proactively.
Partner onboarding
Onboarding routing partners is a potential drain on operations, taking personnel and time away from the business for what is essentially a data-gathering task. Onboarding partners’ pricelists is an essential task – and must be done with precision.
If pricelist and invoicing templates or currencies are set up incorrectly, this can create significant downstream impacts. Doing these tasks manually is therefore time consuming, comes with a high risk of error, and can be redundant if historical data already exists for this partner.
AI is a welcome relief to this process. It can not only be used to automate these manual processes – reducing errors and time spent – but predictive analysis can also support with setting up templates where historical data is available. Explore our iXLink product for automating price list management
Support services
Finally, customer service is a popular area for AI across telco functions. OBH is no different. Chatbots are getting more and more sophisticated, shortening response times by using conversational AI models based on templated SLAs to respond to queries and resolve issues.
Better yet, they have the power to learn. By solving an 'unprecedented’ problem for one customer, chatbots could evolve and use this knowledge to help the next. Customers would no longer need to sift through years of posts on community forums for someone who has experienced the same issue as them – they could just ask AI.
What’s next for AI in OBH?
These four use cases will be the focus for the near future – especially fraud management. The next few years will see these four use cases grow in popularity, but we’ll also see AI capabilities develop. Here are the two key areas the market needs to focus on to empower AI’s growth:
1) Investment
One potential barrier to AI development is investment. Today’s ‘simpler’ versions of AI tools – i.e. automation and analysis – are delivering great results, and the ROI of more complex tools is, relatively, unknown. In a time when optimizing margins is essential to survival, it’s understandable that investing in next-gen AI isn’t everyone’s top priority.
However, convergence could make AI investment more accessible.
For example, we at Telarix are investing in solutions that support multiple wholesale services, including not just OBH, but also SMS, A2P, RBM and more. This means that investment into truly AI-driven solutions will have greater applicability across the business and a greater ROI.
2) Data, training & reliability
As we’ve discussed, today’s tools are focused mostly on productivity, not decision-making. Mostly, people are reluctant to hand over their trust. In many cases, organizations still aren’t using automation to its full potential because of questions about reliability or because they can’t trust a black box.
However, with more training and validation they will become more reliable, consistent and trustworthy. It just requires time. For example, Telarix has been embedding rules into our traditional automation models for years, which means we can trust our solutions with greater levels of automation today.
3) Trust and decision-making
Finally, some areas of OBH will always have human intervention built in.
For example, look at dynamic origin-based routing. There’s simply too fine a line between reducing operational overheads and one false positive causing a costly network failure for this service to be entrusted wholly to AI.
Telecoms have taken their lessons from Google and Microsoft’s own outages in 2025 and 2024, respectively. If a simple software update can fell these global giants, then vendors will ensure that the most important trading areas are unaffected by AI until it’s bulletproof, and even then, will remain under human supervision.
In summary, AI in telecoms today remains (successfully) focused on operational efficiencies. The next few years will undoubtedly see these use cases grow, thanks to more training, testing, which will build trust. Nevertheless, no matter how powerful AI becomes, and despite some people’s hopes for a fully automated telecoms world, humans will always be kept in the loop for the most business-critical applications.
Interested in Learning More?
Connect with a Telarix expert today to learn more about our various offerings.