How Southern Company turned governed data into enterprise AI value

In this episode of Energy Transition Talks, CGI’s Peter Warren speaks with Joyce Solomon, Data Analytics Manager for AMI at Southern Company, and Doug Leal, Vice-President, Data Analytics and AI at CGI, about how utilities can turn trusted, governed data into scalable enterprise AI value.

The conversation explores how Southern Company is using advanced metering infrastructure data, cloud-based architecture, strong data governance and business-led innovation to improve grid operations, customer engagement and AI adoption. Joyce and Doug share real-world examples, including meter-to-transformer validation, revenue recovery, EV charging insights, HVAC efficiency detection and the use of AI-ready data foundations for RAG, large language models and AI agents. 

Listeners will learn why governed data is not a barrier to innovation, but a critical accelerator for trusted analytics, scalable AI and enterprise-wide value creation in the energy and utilities sector.

Visit our  Energy Transition Talks page

Transcript

Welcome And Guest Introductions

Peter Warren 0:06

Hello everyone and welcome back to our ongoing series on the ever-changing world of energy and the transition and how things are moving forward. Got a great session for you today. I've got a wonderful couple of guests. I've got Joyce Solomon, the data analyst manager for AMI for Southern Company, and also Doug Leal. Doug's joined us before, our VP from uh Consulting for Data Analytics. Uh Hi, Joyce. Do you want to say a few words and introduce yourself further?

Joyce Solomon 0:33

Yes, thank you, Peter. Hi, everyone. I'm Joyce Solomon, the data analytics manager in the AMI space in Southern Company. Have the privilege of managing 4.4 million meters across Southern Company. So very excited to be here.

Doug Leal 0:48

Excellent. And Doug, yourself. Hello, everyone. Yes, Doug Leo. As Peter mentioned, I'm a vice president of data analytics and AI here at CGI, helping our clients to drive the most value out of their data. And I've been doing data projects and machine learning before machine learning was, you know, on the mainstream, before machine learning was sex, if you will, for uh for many years. And I'm very happy to be here and looking forward to the conversation.

Vision For A Practical Data Strategy

Peter Warren 1:15

Oh, that's wonderful. And I often forget to introduce myself. I'm Peter Warren. I'm the global industry for energy and utilities here at CGI. Um, so Joyce and Doug, you've uh done a bunch of different things together presentation-wise recently because of the successes you're having. Uh recently, earlier this year, you were at uh Distributech in the United States. You did a great speech from the presentation uh up on stage, and that received a lot of kudos. Uh, but you also just recorded a uh podcast with Energy Central. Uh Doug, do you want to maybe recap what that is so people can maybe seek that out as well?

Doug Leal 1:46

Yes, yes, absolutely. Yeah, you know, it was always great to uh get together with Joyce, right, and talk you know about our projects and talk about our initiatives. Uh on that podcast, we talk a lot about the goal and the vision for a solid data strategy, right, to get the most value of our data. So uh we dive into the details around uh the focus of the data platform that we start building with you know Joyce's team, which was all around bringing the data close to the business, right, and driving value out of that data. Uh, we dove into uh data governance, right? Everyone's favorite topic. So we you know we dove into data governance, but effectively how to enable data governance as an accelerator, right, instead of you know uh uh break, right, to the entire process or you know, stopping the you know the uh entire process. Uh, we talk about the federated model, how you know we enable instead of one monolithic big data enterprise platform, how we follow more of this mesh approach, which is domain-based uh data platforms and aligning the data platforms with data products, right? Those data products focus on uh providing out the information for that domain to consumers of that uh data, right? When I say consumers, it could be a team, could be a machine learning model, could be a Power BI report or you know, a tableau report, whatever it is, right? Whoever is consuming that data to have all that information in one place. Um, I will hand over to Joyce here too, you know, for her input of what else uh you know we talk about during that 30 minutes podcast that we did in the past.

Joyce Solomon 3:33

Yeah, not just did we talk about the foundation that was built with the data, but also on all the um insights, the value proposition, the AI niche that we were able to accomplish and we continue to accomplish is what we spoke about. Interesting enough, one of the questions that was asked was around how AI is received in Southern Company. I remember us having a detailed discussion on that, and I'm pretty sure every every organization today that touches AI, uh, that's the topic of the matter, actually. So happy to participate and uh happy to talk about that here too today, if that's uh uh a topic of discussion.

Peter Warren 4:16

Each year CGI does a voice of the client survey and we talk to our customers. And I would say Southern Company fits into the category that is succeeding with their investment in AI and their investment in data, they're doing things correct. It was interesting that there's a common characteristic, and only about 30% uh of the companies are doing this, and so you're in an elite group, and it was around the culture, it was around having a platform, it was about having a standard strategy, it was also support of the executives. Um, so one of the things I I really enjoyed when, after hearing all the technical things you did, was hearing how the culture of Southern Company is working together and receiving this.

Leadership Support And Culture Shift

Peter Warren 4:55

Um, why don't we just dive into that, Joyce? Uh what's your thoughts on all of that? How is how's your leadership supporting you? How is uh how are your users supporting you?

Joyce Solomon 5:03

Yeah, uh Peter, I have to say being in this role excites me the most because I see a lot of support coming from the leadership that I report to. They have the same sentiment as me and my organization on moving the niche together with you know turning data into insightful uh insights and information that could be given to other business units or directly to the customer as well. So um, with the journey that we have made, uh, it has been such a great journey because not only is the leadership behind us, but we've partnered very well with the technology organization that we have internally, as well as vendor partners like CGI today that has helped us build that uh fundamental architecture that keeps enabling us to quickly, fast, and nimbly bringing up, you know, value proposition and use cases to light. And of course, you know, any business unit that we serve from the AMI perspective, that's just a rule of thumb that I have that I always like to have champions joining us in the journey because you know, they are the SMEs in their area, they will know what value proposition each of the use cases that we can deliver or we will deliver and how they will use it. So I think Peter, you know, putting all of that together in a circle is how it's I've seen great success coming out of my organization.

Peter Warren 6:34

I love that because it's less about the technology and more about the organization, and uh, we do see that as a big thing. Uh Doug, uh, how would you like to comment on sort of how you've been working with the organization? And uh, I mean, you you walk a line between the IT department, you walk a line between the lines of business. How does that all work?

Doug Leal 6:51

Yeah, yeah, it is it is um very interesting dynamics because uh we need to make sure that we are following all the guidelines, procedures, and standards from under the uh IT organization, uh while making sure we are delivering you know value right to business teams. Uh, but I think both organizations, right, they are aligned on the purpose, right? Uh there is a very clear signal, right? When the conversation starts, either if it is on the business side or if it is with the technology organization, when the conversation starts, it always starts on the enabling business decisions, right? It never is about a tool, it never is about how we're gonna use this new uh technology, right? It is all about the business decision, right? How are we gonna enable the business to make better decisions? How are we gonna bring the data close to the business so they can innovate? And that has been a very rewarding process, right? Not to focus on the technology, but leverage it the technology to solve business problems with IT supporting us during the entire process.

Peter Warren 7:59

And and Joyce, uh you know, that I talk a lot to the 70% that aren't getting the success. So, what's what's the secret recipe between your organization? I know you don't just have a CIO, you have a CITO, that's a factor. Uh, but even the definition of your role is a little bit different about uh data and who has it. I don't believe you have a lot of silos or problems with silos.

Breaking Silos With Shared Data Products

Joyce Solomon 8:20

I think we did have that, Peter. We have to acknowledge that, right? At the beginning of the journey, there's been a lot of silos. Data has been in all different locations, you know, accountability was only on their own data set. I think it's the barrier of including technology tools and enablement and also being a partner to journey, not to take over, was the culture shift for every organization to come together and put the data in one location, that is the cloud environment that we have today, but also to help each other and service each other off the data products, and that helped to build a clearer role around the stewardship of data, how we put governance around it, and then you know, the first use case that was built and how it showed value instantaneously. Everybody wanted to jump in, everybody, everybody wanted to use the data, everyone, everybody wanted to be part of that journey. So I think it was an awesome journey that we took uh with the um reinstatement that you know the tool, the data, the insights is just not owned by one organization. It is shared across for many different organization business units to see value.

Peter Warren 9:44

I love that. I I see that as the one of the one of the many benefits you had. Uh, are there some use cases you two would like to highlight uh that come to

Transformer Mapping With AMI Analytics

Peter Warren 9:53

mind?

Joyce Solomon 9:53

Yes, absolutely. Believe it or not, um I I am so privileged to handle the AMI data that comes out of all our smart meters, right? I will go anywhere, and today I would want to say that AMI data is the most sorted after data and is the most sexiest data. So I feel very empowered having uh that data set, owning it and governing it, and then providing it as info uh insights to others, right? So the first use case that I was challenged on building upon in the AMI data was, you know, doing the art of possible on identifying how a meter is connected correctly to a transformer or not. You know, you know, when you look at that use case, you might think, okay, this is not really sexy, it's not user appealing, but it's actually fundamental for the grid's health, right? You gotta connect the right types of meters, the right size, the right amount of meters to the transformers so that the upwards load is well balanced and your grid becomes healthy. So we use the art of data science to do that. Not only did we use voltage uh signatures for this, but we were able to identify missing meters. And then uh with the uh insights that we delivered, we also were able to do recommendations. What is the next best transformer this meter should have been connected to? So those kind of corrections were done, you know, some automation went into that for self-correction as well. And then, you know, um that use case became so popular in the power delivery space, and they were like, What else, Joyce? What else can you do? What else are you seeing out of this, right? When we

Finding Energy Theft Through Anomalies

Joyce Solomon 11:40

were doing that use case, um, we were able to discover a new use case, which was usage death. Um, believe it or not, as the models were running, we were, you know, we were finding out um anomalies on meters that are still drawing usage, but they're not tied to our account. So we were able to then do uh triangulation uh architecture using data science, and we were able to spot those meters that is considered as loss meters because energy theft is taking place, and for every operating company, we were able to bring back revenue. So, you know, that's you know, one or two use cases, and it continued

EV Charging And HVAC Insights

Joyce Solomon 12:23

to snowball. A couple of months later, we were able to use just AMI data, and we were able to detect EV charging at home and the charging behavior, the time of charging, and that helped to reach out to customers proactively in communication and you know, educating them on becoming a new EV user and the rates that we have, the programs that we have, because you know that ties directly to customer satisfaction and also keeping our grid healthy so that we educate them not to charge their uh EVs at home on the peak hours as well, right? And I can go on and on, Peter. The next thing that we discovered from that was um, you know, HVAG uh HVAG B heating or cooling and the inefficient of HVAG and its energy scoring as well. I mean, that caught the eye of our leadership in the company, and today we were able to use these as our insights as to how we can target customer satisfaction, um, you know, keeping our grid ready and healthy as well.

The Popcorn Effect Of Scaling AI

Peter Warren 13:35

I think I when we we we started talking about this with Doug and we were and yourself initially, it uh and other people have coined it the popcorn effect. One group had success, then another group said, Oh, I want that. So change management of these people coming towards, and you know, often there's a push-pull between the lines of business and the IT department, everything. But because they wanted the outcome, they adopted the IT rules. Uh, Doug, do you want to maybe talk about that push-pull that created this popcorn effect?

Doug Leal 14:02

Yes, yes, absolutely. And and it all comes down to the to the company culture, right, which is to enable the business to to uh make decisions, right, uh based on the data, right? And they are the subject matter experts of that domain, right? So giving uh the team access to the data, giving the business team access to the data uh that they can trust and innovate. And uh the tool set, right, that enables them to implement the uh to leverage the large uh and and uh large language models and the greatest and latest versions of you know large language models out there, uh that was the key differentiator, right, to make teams innovate, to make teams you know deliver value. And uh what Joyce just described here, right, that's the long-term uh benefit, right? Which is this whole compounding value of solutions being delivered. So uh once you know, teams, right, once business teams they have access to trusted data, uh, with this shared model you know, uh uh environment that we you know establish, the second and the third use cases become faster than the first, right? And that's when you start seeing the enterprise value, right? The enterprise momentum, right? Going back to your point, Peter, the popcorn effect, right? Where teams see, like, oh yes, uh that was successful, I have a similar use case, let me get on boarded to this platform and leverage all of the infrastructure, all of the automation that was already established, uh, for those teams to just focus on the use case, right? They don't need to worry about infrastructure as code because that problem was already solved. They just come to the platform and focus on the use case, which I say only on the use case, like it's something simple. It is not, right? There's a lot of complexity, but uh as much as we can remove from you know, from from from the business plate in terms of you know, you don't need to worry about deployment because that's already figured out, we already solved that, the better they will be able to, you know, move forward and move fast, right?

Peter Warren 16:11

Well, that's brilliant.

RAG LLMs And Guardrails

Peter Warren 16:12

Well, at as we start to wrap up here, Joyce, uh maybe I can give it back to you for a few minutes here just to talk about that culture of innovation that's happening. So it really moved from a culture of uh command and control to a culture of innovation. How how do you how is that working and how is that going for you?

Joyce Solomon 16:27

It's doing great, actually, right? Um, Peter, because uh not only do I get to innovate, not only have I been given the space to do so, I have been successfully influencing other leaders like myself in the organization to move forward and you know to be able to produce and be at the niche. So today, you know, with the tools that we have, just like what Doug said, we're moved into the space of RAG, LLM, AI agent deployment. And of course, with all of this comes the safeguard of our data as well, right? We have data privacy that we have to handle, governance that we have to handle, and also um, you know, the leadership, uh having their trust on uh leaders like me to deliver those things for the betterment of the company to make decisions as well. So I

Wrap Up And Final Takeaways

Joyce Solomon 17:24

I'm in a great spot. I enjoy my job and I want to continue to influence others to do the same.

Peter Warren 17:30

Well, that's brilliant. Maybe we'll end on that wonderful point how you guys have taken uh actually what people would think is governance as being something as a block and made it a strength, a power, a foundation to move things forward. Um it's a very inspiring story, and thank you very much for saying it. So uh Doug, uh give you a goodbye and Joyce a goodbye, and we'll we'll talk to you everybody on the audience another time.

Doug Leal 17:52

Well, uh thanks everyone. It's been a pleasure to be here. Uh great conversation as always. Uh, it is a great time to be working with data, great time to be working with AI. Uh so looking forward to the next you know 10, 20, 30 years. So it's a very exciting time.

Joyce Solomon 18:07

Thank you all for having me and you know, happy to participate with CGI. Uh CGI has has been a great partner, collaborator for us. So we continue to look forward for this uh partnership and for us to innovate together. So thank you.