Unlock Growth and Fuel Efficiencies as an Agentic Telecom | Dreamforce 2025

It won't be a Salesforce event if we don't. Thank you guys for being here. Thank you all for being wonderful customers and partners. And thank you all for being here. And thank you all for being here so early in the morning. So really appreciate it. My name is Kishore Sunedanam. I'm the global industry strategist for telecommunications. And I'm happy to be here talking to you guys this morning. I just wanted to set the stage for what we're seeing in the industry. The stat on the right hand side, the 2.9% is the annual telecom growth, according to a study that PwC published at the beginning of this year. And earlier this week, I asked our good friend chatgpt, just with the publicly available information where telecoms are trending globally. And I'm pleasantly surprised and happy to state that most of you guys are doing better than this. So that's great. And earlier this summer, we actually did a survey of the top telecom executives globally to understand what are the top challenges. And as you could see on the screen, increased competition, effective usage of data, and then productivity. It's nothing new. It's the same thing that we've seen year over year. There have been multiple transformation projects, there have been multiple initiatives to get that down. But I think for the first time there is a glimmer of hope and a sense of progress, if you will, in this arena where generative AI, I think everyone's thinking, will actually help us and some of the early results we're seeing, and then you've been hearing throughout the conference, and then you'll hear from my good friend Tiago in a minute that it's actually working. And when we asked them, hey, what are the areas that you're looking for, potential benefits coming out of this? No surprise. It's really about how can we be more productive, how can we be more efficient so that we can actually pursue new business models or new revenue strategies or new channels? How do we get access to data at scale? How do we democratize data? Until LLMs came around, the insights to data would have meant an army of data engineers working at a particular problem. But now we have the ability for everyone to get at it. And then the last thing is really increasing the employee capacity by augmenting them with digital workers. But even with that backdrop, what was surprising to US was only 48% of the survey results have indicated, or people have indicated that they're actually using generative AI technologies across their processes. Still good. Half of you guys are still in the trial and experimentation phase. So to that end, we're introducing agent 4C360 for communications so we can help you get started on that journey faster without industry depth. For the industry context, we'll be providing out of the box data, metadata and semantic models. We'll be shipping pre configured workflows and automation technologies so you can accomplish complex tasks. And we're also providing the ability to interrupt between Salesforce applications and your tech stack leveraging standards. But with that, rather than me talking, I want to introduce you to Tiago Loira who leads the transformation office at Nost Portugal to talk about how they're going through this journey. And then he's got a really interesting demo at the end. So I really urge you guys to stay till the end to watch it. Thank you.

Okay, thank you. Kishore. Hi. Hi everybody. Thank you for being here again to listen to what I have to share with you. And thank you Salesforce, to give us the opportunity to to be here sharing our experience with all of you. We are a telco company in Portugal. Just to give you an overview. We are on the fixed and mobile businesses. We have a good coverage and a good customer base in Portugal to have reference for these numbers. We are about 11 million people and 6 million homes. So we have a good coverage. We also are in the media business. We produce content and we have a lot of movie theaters in Portugal. And we are trying to expand. As we all hear, the telcos are having a lot of challenges on creating new businesses. So we are trying to launch new businesses regarding insurance, regarding home security, regarding smart homes. So we are trying to diversify all our businesses so that we can have a broader scope of customers. The demo that I will show and the use case that I will bring to you today, it is related to Wu Woo. It's our low cost brand for mobile and fixed. And it was based on customer centricity and on four pillars. The speed. It was very important to be a very simple and agile experience for the customer based on quality. We are trying always to improve the customer experience and the functionalities and capabilities that the customer has to do. In the Woo app, the efficiency because being a low cost brand, it is important to slim down the operations and optimize the operation. So it was a goal to have a lot of efficiency on this new brand and innovation because being digital, we are always trying to to bring new features and new functionalities for our customers. The challenge for us it was to be 100% digital. So we want to be completely out of the regular store Use cases. So we don't have stores, we just have some kiosks. It always happens. Everything on the app but the customer experience. We think that it's very straightforward. If you have a mobile phone with any SIM, you can have a fully 100% digital experience in a couple of minutes. You can have your subscription running on your phone, on your home without going anywhere, without going to the store, without any trouble to do it. Also, and this is a very important goal that we achieved. We were leaders in satisfaction with an NPS of 59. As I mentioned, it was very important to us to be customer centric and this was one of the goals from the beginning. And I think that the experience that we put in place and all the offers that we have lead to this score. But what bring us here is agentforce and it is the service model that we are using on the Woo app. We have a chat that is 24,7 available on the app. I will explain a little bit more about the roadmap that we did, but the goal, it was to have the most of the service. We would like to be 100%. It will be not possible or not, but we want it to be 100% human, free of handovers. So we look at Agent Force and we are developing right now and testing it because it's in the testing phase, the integration between the Agent Force and our current sheds, we need to integrate it with our app. We are trying to use the invocable actions as an accelerator for the overall experience. We are trying to leverage on the data that we have on the platform and we are trying to reduce the latency that we have because right now we are using APIs and with agent Force we foresee that it will be more quick to access all the customer data. I didn't mention, but we are not all in on Salesforce and we what I mean by this is that we have product catalog, we have cpq, we have the customer assets and order capture and order management. But we don't have service and we don't have for instance the contacts and CRM from the customer. The Woo evolution, the brand was launched on 2020 and as I mentioned we started with 100% human assisted. So. So every interaction it was hand over to a human in a low cost brand and in a scenario where you want to reduce the cost, it is not feasible. But we start with a lower customer base. We started scaling in 2022 and with this escalation of the customer base, we feel the need to do a little more automation on the customer service. So we created a flow based chatbot that diverted 45% of the endeavors to the human agent. The next level it was to include Generative AI and Energentic platform. In our case, we started with servicing and with this step we were able to even reduce more on 22% the interactions that we have with humans on this scenario. And right now we are finishing the implementation of the AgentIQ for Salesforce. And our goal as I mentioned Is to reach 90% of call deflection without eating the commercial ratios and keeping the customer satisfaction level. As we are seeing this, we are having a 4.515% rating on the responses of our agents. So we want to keep this and to keep going forward to our goal. That is to have no human interaction in all this platform. And how did we do it? I will not explain in detail this slide in the sake of time, but we had some challenges. We have always new technology appears A2AMCP, new LLM providers, new solutions like Agent Force and how to integrate them. We need to integrate this solution into in our existing touch points, in this case the chat that we already have. And we need to do a production grade solution addressing the privacy, the security, the omnichannel and so on. So we created an AgentIQ platform. As I mentioned again, we are not all in on Salesforce that is based on five function agents. We have a rag agent for fax that is responding to the common questions, the most common questions that the customer has. We have two agents, the mobile agent and the subscription agent that are tool based agents that using APIs are getting the data from the subscription and from the mobile service and answering all the servicing questions regarding this. And we added the Agent Force, the Agent Force Sales that is based on the Agent Force that is checking the catalog and answering all the offer related questions. And for sales and in all these interactions, what we can have is a change of context. We can use several of these subagents in different approaches to answer different questions. And we have started on the build part. For now we are considering maintaining the build part because as I mentioned, we are not all in on Salesforce but we consider that Agent Force because we have the catalog and the CPQ and the assets on our core in Salesforce. The Agent Force makes a lot of sense for this. How did we do it? We partnered with Accenture that is our primary implementation partner that did the first implementation of the Woo app and the U experience. It was based on the traditional Omni Out Omniscript integration procedures solution. The traditional scenario. And then we evolved to this chat approach where all the interactions are chat based. For now, we are using custom actions for leverage on the logic, on the business logic that we already had inside the omniscript. We are leveraging on our APEX class and our integration procedures that we already had to create the custom actions and the topics that the agent is addressing. And now we are testing the invocable actions that we consider that will be an accelerator for delivering more use cases on this platform. What was the benefits that we took until now from here? First we consider that customer interaction time, it's decreased by 40%. So going through the process that we already considered, that was simple. We measured the experience time and doing it on the conversional platform, it's 40% lower than we see on the OmniScripts. Also, we streamlined our buyer experience and reduced the potential risk of having issues in the integrations, in the APIs and so on by 60% because the logic is inside the integration procedure and the APEX classes and it is streamlined to that. And not least, but not less relevant the time to market. We took 75% less time to develop the conversational use case versus the implementation of the OmniScript. But as Kishar was mentioning, I have a demo here because probably a picture is better than a thousand words. So on the left side we have the current experience where I interact with the shed and what we did basically here before we have the agent for sales we asked about. Sorry, this is in Portuguese, I will translate. I want to change my tariff plan. And what the chat will respond is to redirect the customer with a dimpled link to the action on the app. But the catalog is messy. They have a lot of options. The customers can be lost inside of it. So we created the new experience that we are seeing on the right that is embedded already on our app. We are not using Gateway yet, we are using rest, but it is embedded on the app. We started with a simple question. What offers do you have available? The agent right now it's going to the catalog and to check what is the customer tariff plan and what are the rules that apply to the tariff plans available. The customer checks one of these options. I want to change for the 5, 5,000, 6,000 minutes offer. And right now it will take about 30 seconds. This is the reality, but it takes 30 seconds. Why? Because we are doing all the validation of the rules, we are doing the customer scoring and then we are creating the actual order for the customer to fulfill their request. And the order is created and the customer has his new tariff plan.

Thank you, Thiago. So before I let you go, I just wanted to highlight a few things you said just for the audience. So one is you were able to reuse your existing assets to build it, right? Cause there was a lot of conversation the last couple of days around tech debt and am I dropping everything off. This technology is moving so quickly. So. So that's one. That's great. The second thing is you are testing the industry agents for all the new cases, so you want to use even more out of the box. So hopefully the dev time will shrink even more than the 74% you talked about. And the last thing is, what we just saw here is not a separate screen, but it's actually running inside your digital native app that you already have.

And I didn't explain that, but all the authentication of the customer, all the authentication of the agent, it's been transparent for the end user.

Excellent. Thank you. So just one other thing for you guys. I talked about a few AI trends and stuff. So if you scan the QR code, you can get the full report, but there's still more content that's very telecom specific in the next today, in the afternoon, tomorrow. So the most importantly, at noon today, we have our keynote. It's on the Intercontinental Level 3 Grand Ballroom. And if you scan the QR code, some of the sessions you may have missed here are already on Salesforce Plus. So with that, thank you, Tiago, and thank you all for being here.

Thank you.