Over the past decade, Experian has been on a journey to transform from a traditional credit reporting company to a platform business. As part of that, it has been investing in technology, AI, and analytics.
One of those investments was creating Experian Assistant, a GenAI-enabled tool that helps financial services companies explore Experian’s treasure trove of data, build and deploy analytical models, monitor their performance, and accelerate new product and service offerings to the market. Earlier this year, Experian Assistant won the 2025 BIG Innovation Award in the financial services category.
Shri Santhanam, Executive Vice President and General Manager of Software, Platforms, and AI for Experian North America, recently spoke to InnoLead about how the credit reporting giant launched Experian Assistant within nine months from idea to market, and the time-savings it is producing for clients.
His Role
Shri Santhanam: My role as leader of the software and platforms business in North America is to help Experian use data, technology, and AI to drive impact to businesses and consumers. I oversee the software and platforms business, and I’m responsible for all aspects of it, including product innovation, growth, and impact on our businesses and consumers. I also wear the hat of the global executive sponsor of our AI agenda. In that role, I help amplify the impact of AI across Experian and bring greater impact to the customer, both consumer and business-facing products, as well as our internal productivity.
The Project
SS: When the ChatGPT Meteor hit, we said the power of unlocking AI is to have 1,000 flowers bloom. So, it wasn’t about getting the two or three smartest people in the company in a conference room and trying to solve this; we wanted to create access and encourage people to understand and educate themselves on this and come up with ideas.
What we found was a couple of data scientists came to us with a really interesting idea. They said our data is extremely valuable and impactful when it’s used, but we have a lot of smart customers who ask us hundreds of questions daily that are very specific and detailed about our data like “Can this information be used in a FCRA-compliant model?” or “What are the attributes associated with delinquency?” Normally, the way we answer these questions is to queue up experts in our business to answer them and, sometimes, it requires research.
We asked ourselves, “What if we could use AI to put a personal expert – with the standard of our best data scientist – with each of our clients 24/7 and indefinitely?” That was our vision. This personal expert could answer all their questions. Then, as we started building out the vision, we said, “Wouldn’t it be amazing if this expert could not only answer their questions but could also give them snippets of code or help them run this on our systems and make it easy?”
…We set some clear ambitions and goals for the big problems that we knew we wanted to solve in a certain timeframe. We said we would stay true to those, and be flexible on the how and the details.
Innovations are never linear. In the end, they might feel linear — but it’s never linear. You’re going down these nooks and crannies and figuring it out. For us, we did a couple of things to manage for that. First, we set some clear ambitions and goals for the big problems that we knew we wanted to solve in a certain timeframe. We said we would stay true to those, and be flexible on the how and the details. The decision we made was that by the annual Vision Conference, we wanted to have at least two or three clients who at that conference were telling a story about how we have made their lives significantly easier and faster, more productive, with something like this. That was our goal. We took a bunch of twists and turns to achieve that goal but, by and large, we hit that goal.
How It Got Green-Lit
SS: Generally, how products get formally signed off is they need an executive sponsor. In this case, it was actually very straightforward. When all of us as leadership saw the goal, it felt like we were driving it in reverse because people were saying how we needed something like this and let me help you, so you can go faster with this.
I think there were the usual questions and guardrails like let’s ensure we’re doing it in a way, which pays the most attention to our data privacy and our responsible AI practices. But in general, there was overwhelming support and recognition that this sort of thing, as an innovative company that we are at Experian, needs to be encouraged and fostered.
Video: “A big part of the challenge we faced here was a lot of this was net new…We were bringing new capabilities with new technology,” Santhamam says in the clip above. “One of the big things that helped was being intentional about getting experts in each of these areas to just sit down and really have an open dialogue, educating and understanding each other…”
The Three Biggest Challenges
SS: One of the big things we wanted in our assistant was a very high degree of fidelity, accuracy, and trust. Getting that right took a fair bit of work in defining a set of skills and a framework and how we would do that. It also required going through the details and figuring out how to ensure that we had the right governance in place. We ended up making a set of investments and creating some novel approaches to do that.
The second area that was challenging was being specific in designing which problems we were solving and where it’s most important for our data scientists. We took an iterative approach and actually got something going very quickly in a matter of days and weeks. We sat down with our clients and what we found was slightly counterintuitive. We found some pretty simple and straightforward problems that were getting good answers for those who found them extremely valuable. Often in innovation, you’re looking for things that may feel like rocket ships launching versus simply like weather balloons flying. Sometimes, the weather balloons are actually really powerful. They may not look glamorous, but they solve important problems.
Another big challenge was we had to move multiple threads forward. We had to develop the product, we had to have an engagement and delivery model with some of our clients and we had to have a thread on risk, responsible AI, and compliance management. One thing we did early in this process was to set up a Global AI Risk Council, which helped us fast-track some of these things. It created swim lanes because if I’m a product person and innovating, I want to primarily focus on customer problems, how to build it, and how to solve it. I would want to abstract questions of compliance and risk and provide a way to support them versus being consumed by all of that.
The risk council allowed us in a very clear and streamlined way to provide some guidance in those decisions and it helped inform our product vision early. It kept us focused and doing that effectively means you avoid churn and burnout with the teams. It also allows the teams to understand the choices you can make and be intentional and clear on those sorts of choices.
Metrics We’re Tracking
SS: We are basing our metrics largely on client feedback, client engagement, and client traction because, ultimately, we’re looking at the impact it’s having on our clients. It’s still early days, given when we rolled it out, but we’re getting extremely positive feedback from our clients. Some of the KPIs and metrics we’re looking at are how productive is the assistant for our client users and how is it helping them. We’re seeing good results there. We’re finding that on average, some high-value analytical tasks are getting 60 percent to 70 percent savings, in terms of time. We’re also seeing good engagement and a lot of demand and excitement from this. So, we’re seeing positive results.
