- Tech Mahindra is helping telcos all around the world to incorporate AI into their customer service
- This is especially important for a large percentage of calls that fall into two buckets
- A Tech Mahindra executive said customers just want to solve their problem fast
DTW IGNITE 2026, COPENHAGEN, DENMARK – Companies such as Tech Mahindra enjoy TM Forum’s annual conference because it brings together their key constituency: telco operators who need help with their operation support systems and billing support systems (OSS/BSS). And while OSS/BSS has never been a particularly sexy topic, it’s gotten far more interesting with the advent of AI. In fact, customer experience is probably the number one thing that operators are using AI for.
Manish Mangal, president of America’s Communications Business for Tech Mahindra, said customer care typically falls into two buckets. The first is “containment.” This refers to customers who are calling with a problem that is fairly straightforward and can be answered by a chatbot or conversational AI. The second bucket is “proactive,” where the customer needs human assistance.
In terms of containment, people often call because they have a billing issue or because their service is degrading. “You can actually get a lot of those things addressed through either the chat or a conversational AI agent,” said Mangal. “Chat was much more popular, but now, the conversational AI is allowing a conversation to happen in a very fast way without having a human involved.”
Of course, the benefits are better standardization and less cost.
But sometimes, a proactive response is necessary. “When you have a customer who is not the happiest customer, you don't want to give them some artificial thing. That's when you get a human involved to make sure that you are emotionally connecting with them, and you are helping them understand how to solve their problem.”
Amdocs is doing some really cool things with customizing AI agents so they relate better to each particular customer. But Mangal said, “There is some value to that, but honestly, it's not something that's going to make or break the behavior of the customer. What's more important is that when the customer reaches out, they want to get done fast.”
But are customer interactions really faster?
There was a joke on LinkedIn about the “pizza paradox.” In 1995 if you wanted to order a pizza, you would call the pizza parlor, and it would arrive in about 30 minutes. In 2026, you have to go through so many steps: find the website or app; create an account with email verification; add your address with a pin on the map; type in all the selections for your pizza; put in your credit card information. The pizza arrives in 85-140 minutes.
Mangal said the problem with pizza ordering is that there are too many options. Back in the day, people just ordered a cheese or pepperoni pizza. He’s a firm believer that AI will speed things along at telco’s customer service departments
Technical debt
Fierce asked whether telcos are ready for AI given the technical debt they’ve accumulated over the years.
Mangal acknowledged that technical debt is a real problem. Data is residing in multiple platforms in different formats. “You can’t just extract it out and put it somewhere else,” he said. “It's too deeply integrated. That’s where the only option is to have API access, so that you at least can call out the data you need.”
He said telcos are typically following a federated data model, where they create a data structure on top of their existing data sources. This AI-enabled model is also called “bolt-on AI.” Mangal said you can only do AI-native when you have the luxury of building a greenfield network.
He also noted that all telcos have their own data. “It’s not like you can take everybody's data globally together and say, ‘Now, I'm going to build an agent that is ubiquitous.’”
The good news is: there are a lot of learnings and tools coming from the AI world to build AI agents for customer service.
Individualized pricing
Fierce asked if operators are moving toward a world of micro-segmentation, where they customize plans to a very individualized level.
Mangal said, “Remember, I told you the number one reason why customers call is billing? By giving them a lot of differentiated pricing, I think you're inviting more calls on billing.”
He concluded, “The only reason everyone's talking about all of this is because AI can now do that. It can take all that data and segment all this stuff to the point that it may segment a little bit too much for people's comfort.”