Ray Le Maistre, TelecomTV (00:13):
Two of the biggest technology topics in the telecom sector right now are agentic AI and network APIs. I'm here today to talk about both of those with Tapas Ranjan, VP of Product Management at Rakuten Symphony. Tapas, great to see you. Thanks very much for joining us today. Now there's a lot of focus right now on the potential of agentic AI in telecom, but do AI agents represent a net positive shift for the industry or do the risks associated with the likes of security, control and more outweigh the benefits?

Tapas Ranjan, Rakuten Symphony (00:50):
Thanks, Ray. I think for the question. What I feel is AI agents are net positive, but only if we move into the trust-based to policy-based execution. Because how I see agents, it is making a fundamental shift in how we manage the complexity across networks, across enterprises. And I think the industry has been facing this problem statement or struggling with this something called as control paradox where we need to make sure that we need to give the autonomy to the agents. But on the other side, it is very much we have to understand that we are giving control on autonomy to an agent in a non-deterministic behaviour as well. But on the other side, if we see it from a guardrailing point of view that we should not draw a line at the technology front itself, but we should do it at the point of execution.

(01:47):
My view here is that we must implement policy as a code, which can implement the agent as the advisor or the orchestrator. But whenever they are executing something, they are bounded with respect to the policy engine of that network and the operations on how they wanted to approach. I feel that can maintain the separation, but with that, what we will get, we will get the intelligence out of the agent and the platform will act as the safety envelope. And the benefits when we talk about reduced OPEX, we talked about self-healing, when we talked about faster time to market, all of that can outweigh the risks that we've been talking about. So it's more about that, how much we are offering back to the agent. We are not giving a key of the kingdom to the agent itself in order to drive everything, but definitely we are giving them an interface to execute faster, but in a controlled manner so that the positives that we are looking out of these shifts in the market, the AI commencement which has happened in the industry can actually benefit back into the industry as well.

(02:57):
This is what I think about it.

Ray Le Maistre, TelecomTV (02:59):
Okay, excellent. Now obviously this is something you've been looking at quite closely and developing at Rakuten Symphony. So how is Rakuten Symphony leveraging the evolution from APIs to AI-driven agents to transform telecom operations and importantly customer experience?

Tapas Ranjan, Rakuten Symphony (03:22):
I think from day one at Rakuten Symphony, we approached the network as a software platform, not as a hardware assembly that we initiated the network then when we started to build the whole network from day one. We have been leveraging this evolution to transform our platform to be an agent aware ecosystem. And how it is helping us through is that we are moving our engineers from manually changing the APIs to solve a problem with how we used to do on running a MOP or running an automation. Now with the commencement of these MCPs, agent to agent protocols, it has given a fundamental shift on how we used to leverage these APIs back into the system as well. So internally to Rakuten, we are also building domain specific agents, which can consume our service catalogues or the MCPs that we've been creating internally so that it becomes easy for us to do a certain use case.

(04:21):
When we talk about use case, it can be from a small use case to a larger use case if I take an example that every software industry, if I talk about our telecom specifically, goes through a lot of changes every day. And if let's suppose there is a specific issue we see in a network, it becomes very, very difficult whether that particular issue has been caused because of the change happened in the network, yes or no. And in order to do that, traditionally, you might look into your change management process, your configuration management, your telemetry services and all that. What if the agent can actually act on your behalf? That agent can help you correlate your tickets, your changes, your incidents, your alarms all together in one. So the manual work which you are trying to do can be actually subsidised directly by the agent itself.

(05:14):
And that is how it can be leveraged, but it can only be leveraged if your network API is your system itself at the bare bone, the infrastructure layer is fully virtualised, this is integrated as well as doesn't have a vendor lock in on how they want it to approach to a particular problem statement as well. And if I also add to a specific customer experience, I think network has to become more context aware to understand what the problem the user might be facing in so that the agent can perform that right action again in a guardrail manner so that it can help the human as well to take the right decision. And that's how we have been working in Rakuten in order to make sure that whatever we have built so far, we can take that leverage and can offer back to the market as well.

(06:06):
And that is how we have been going on to leverage the realisation of what we call as zero touch operations roadmap and the strategy to move ahead.

Ray Le Maistre, TelecomTV (06:18):
Okay, excellent. Well, that makes an awful lot of sense. And obviously there's a lot that can be learned in the industry from the experience of Rakuten Mobile and its virtualised software defined network. So Tapas has been a pleasure talking with you this morning. Thanks very much for your insights and we look forward to talking with you again soon on TelecomTV. Thank you very much.

Tapas Ranjan, Rakuten Symphony (06:42):
Likewise. Thank you so much.