
AI for Public Good: Building Digital Infrastructure
Insights
- Digital Public Infrastructure (DPI) provides the foundation for trust, equity, and efficiency in government services, with agentic AI acting as a force multiplier.
- Public enterprises must balance innovation with governance, ensuring interoperability and responsible AI adoption across industries and borders.
- Agentic AI enables governments to deliver services at scale while personalizing citizen experiences, reshaping how essential services are accessed and consumed.
In this interview, Dylan Cosper from the Infosys Knowledge Institute, speaks with Lax Gopisetty, CEO of Infosys Public Services, about how digital public infrastructure (DPI) and agentic AI are transforming government services through interoperability, governance, equity, and large-scale efficiency.
Explore more videos:
Dylan Cosper:
Welcome to the AI for Public Good series from the Infosys Knowledge Institute.
I'm Dylan Cosper. In this series, we explore how governments are using AI to build trust, boost efficiency, and deliver better services.
At the core of this shift is digital public infrastructure. Think digital roads and bridges for data, services, and secure interactions. And now, with agentic AI, DPI is evolving fast, helping departments connect, enabling real-time decisions, and even acting as digital civil servants.
To explore what this means for the future of government, I'm joined by Lax Gopisetty, CEO of Infosys Public Services and a long-time leader in the public sector transformation. Lax, welcome and thank you for joining us.
Lax Gopisetty:
Dylan, thank you for hosting me and great to be interacting on this platform again. And it's such an opportune time on how the agentic AI public enterprise and commercial enterprise and citizen enterprises are coming together to navigate this uncharted. Looking forward to the conversation. Great to be here.
Dylan Cosper:
As am I. Thank you for joining, really. So I'd like to hop in here and just give a little bit more information about DPI. So DPI isn't just about technology. It involves governance, collaboration, security, and most importantly, equity. What are the core components of DPI? And then also, how is agentic AI helping governments better integrate these areas?
Lax Gopisetty:
As we come into the digital public infrastructure, we see six components, majorly six dimensions which play into this. It's about human identity and what you can do with that identity dimension. It's about economic exchanges, which includes payments and all the exchanges which happens for trade and trade executions. Third is supporting infrastructure and technology which supports this exchanges. And then there is an element of interoperability, which is information exchange, which becomes a crucial element to make sure that it provides a lot more amplified use cases to produce. Now, because of the potential which exists in the digital public infrastructure and what it can do, the governance and regulations to how it can be adopted and deployed becomes super important. The last element is about workforce. Workforce and citizens itself. So I call them as a prosumers, people who produce, contribute into this DPI evolution, and also consume DPI benefits. So in a way, it is a, you, all of us, you, me, all of us are behaving as a prosumers where we contribute into the evolution of technology, and we also consume the benefits of it. Earlier it used to be, you know, somebody would have innovated and somebody else would have been consuming. There was a two different constituents. Today, most of the constituents are pretty much the same. So these six dimensions make our digital public infrastructure as a theme. And it has gotten to a foundational element right now. And we see a huge amount of maturity evolution in that space moving into a more integrated fashion, more adaptive and inclusive fashion as it matures. That's where we see a huge potential to be unleashed.
The public enterprise is going to, almost like they are the custodians of making sure this is deployed and adopted in an ethical way, responsible way, the vulnerabilities are protected in conjunction with making sure the potential which is there in this, the capability and the potential which is there in this new digital economy with agentic AI is realized as well for the better society, better human life. There's so many adverse conditions today which can be addressed much better scenario with the technology evolution. So I see this as so much potential to go after the new areas which we are not going after so far. And really the public sector is truly the shepherd for this. It's providing the guardrails, ensuring that people aren't being left behind, ensuring as people come along that they're treated properly. And what's actually implemented is for the betterment of people, not to the detriment.
Dylan Cosper:
Absolutely.
Lax Gopisetty:
And it's not a new thing, which has happened when public enterprise has defined internet standards. So they have defined nuclear material standards. So they have defined mobility standards. So there is a governance which has always been owned by public enterprise. But, earlier innovation has given a certain amount of time period to evolve. In the AI, the speed at which it needs to happen is what is influencing everybody to think more rapidly than earlier. So one of the opportunities which public enterprise has in this current version of innovation is there is a possibility of co-creation.
Now we are moving from an automation to autonomous city to autonomy operating models. And this evolution is significantly devised by the data and how the learning happens, both humans and machines. We are all going to have a lot of colleagues who are digital colleagues compared to the human colleagues. And we have got to get used to that. So in automation, we were trying to train the stuff with machines with a very segmented data, hence automation could be done. When you are getting into an autonomous city model where there is lot of interoperability becomes very important to bring that across. So if I have to use the previous example, how would an auto manufacturer to a payment infrastructure from a banking side and a road infrastructure can come together to make that possible and how much that can be really useful for safety and security of the transportation. How could I avoid accidents in the future? How could we benefit much more governed transportation or safe transportation? For that to be unleashed, this interoperability becomes, then it becomes automatic. So that's the way we see. And the same issues are there in health care space. Same are there in even just from a purely workforce learning perspective. How could we make sure self-learning is happening today at an individual level? But how could we make sure this becomes a combined perspective-based learning so all of us learn from multiple experiences, which typically happens in your cohorts in the colleges? But how is it happening in the enterprise scale and in public enterprise more specifically where there's quite a bit of siloed operating models historically.
Dylan Cosper:
Very true. So, Lax, how does agentic AI enhance the key components of DPI and accelerate public service delivery?
Lax Gopisetty:
So Public service is, if I define in my lens, it is at scale because every public enterprise in the world probably are the large, they deliver the largest amount of services. It is essential. It is not necessarily nice to have types. It is essential. So there is a, most of them are mission critical. They need to be timely. They need to be at speed because of the scale and importance of them. Agentic AI will help immensely on making sure the quality of service is delivered. So there's no leakages, there's no gap. So citizens get right services at the right time. It also is scalable. Economies of scale is so huge with agentic AI, so it can proliferate at scale. The key is while it does it at scale, it also can be very personalized. That's a key enabler.
Dylan Cosper:
Sounds like agentic AI can play a role in ensuring that every service that a government offers meets the needs of all of its citizens.
Lax Gopisetty:
Absolutely, it does. I mean last week issue within Texas on the flights, maybe the weather channels would have given a broader information, but it needed probably a lot more precision information to the targeted audience. And this technology multiplier, that's where I would say force multiplier. Technology will be at scale, but its consumption will become very pointed. I mean, we are all used to the social media today, and social media operates that way. Or the new economic models of whether Airbnbs or Ubers of the world, all of our perplexities, all of them work that way. The technology is at scale. It harnesses the information and intelligence through a wide variety of sources, but it condenses and delivers to what you want. And if you use a Perplexity example. It consumes from wide variety, but it gives you what you want, or more precisely than what otherwise we would have got. That's the way we should think about essential services. When I'm in a road accident or when I'm in a need of something or when I have a question on my tax refunds, how can I get very precise advice and input for me to take care of that situation? That's where the technology will help to make the entire experience layer to multi-fold much better than current state. As opposed to the frequently asked questions, the generic list of things.
Dylan Cosper:
No, that’s interesting. I want to go back and see, how did DPI and agentic AI improve efficiency and reduce costs?
Lax Gopisetty:
So yeah, good connection. So DPI is a fundamental infrastructure component. So we talked about identity of a human and KYC. So one, you will get an authenticated. So the service gets rendered to the right person. And agentic AI can complete the task or be there to do the KYC. Then if you need to exchange payments, which is the financial exchange or economic exchange which happens in our lives, that is, you know, today agenic AIs are able to complete end-to-end transactions which we might have, you know, got used to booking movie tickets on Alexa or Google Home or, you know, or our city, right. Those are all automated workflows which are able to do it and behind the scenes agents will do it and today they are only doing happy path scenario with less complexity, agentic AI and agents will come start doing more complex transactions that way.
Dylan Cosper:
Could you talk a little bit about how DPI can play a role in that governance for enterprises, for consumers, for citizens?
Lax Gopisetty:
Absolutely. So it's actually very interesting that typically enterprises want deregulations so they can prosper. This is a good situation to be in the reverse order to say, industry is asking for more regulations because they see more vulnerabilities and more risk compared to what more possibilities as of today. But I don't think this is a very unusual scenario. As I said, there were more regulations on internet consumption, there are more regulations on internet payment models and mobile devices. So lot of standards and governance have evolved. I'm sure this will also come into the play. The infrastructure will help to better govern it, create better policies for inter-exchanges. So today, the CIOs and CXOs are not really concerned within the enterprise issues. They are not asking for regulations for the transactions within the enterprise. They are asking for regulations when they have to cross the enterprises, cross the boundaries, cross the geographical boundaries, cross trade commissions. Now today, this is very fragmented as AI is evolving into it.
Dylan Cosper:
We were discussing earlier, in the last three weeks, there's been different states within the US who've enacted and passed their own AI legislature.
Lax Gopisetty:
Exactly. So we have all of these standards that are somewhat non-standard because they're done at the state level at this point. So everything is moving. You pointed out, so Texas released about two weeks back. California has released it. Colorado has released it. And so we'll be there for all agencies. In fact, Defense has released their own LLM now. And because of that, they are releasing their own governing of how this need to be consumed. I go back to the same. Because innovation, production, and consumption happening at that level, which is at the, I would say, bottom of the pyramid types, the standards also will evolve from there and roll up to a national level, roll up to a country, much bigger enterprise levels. But the key is as the infrastructure gets expanded the digital public infrastructure, then the standards will get more proper and evolve into a more standard version. And this happens in any technology innovation.
Dylan Cosper:
As we've been talking, it's kind of got me thinking that a strong digital foundation is key to success with DPI. So how do DPI and agentic AI help speed up modernization and interoperability?
Lax Gopisetty:
Yeah, so as earlier I mentioned about it. So DPI itself as a fundamental core is the exchange, as I said, information exchange in interoperability is the core dimension of public infrastructure. Agentic AI is only going to enhance the speed of establishing that interoperability and using that to create efficient way of delivering the services to the market. It's not like agentic AI is going to transform how we service or how we, the fundamentals of service delivery. But it is going to force multiplier on very cost efficiently, but more precisely, more effectively. I think that's the way I will put that across.
Dylan Cosper:
What are the challenges that you currently see in regard to this like interoperability? What's kind of keeping you up at night as it relates to the challenges related to interoperability?
Lax Gopisetty:
It's a lot of things which are in evolutionary mode. They are not roadblocks, because people believe in it. They need to do. It's a time catch-up. So there is a sensitivity around how the data exchanges can be done. There are guardrails about what information is sensitive and what is not. There are conflicting right guidelines, but conflicting about how much personal information I can share before I can get a personalized service. If I want a personalized service, I need to give my personal information. I'm comfortable or not. So lot of those things have to go through the motions, both at the individual level, at enterprise business units, at government agencies, across the board. So these will come into data exchanges in pockets. But I consider that as an evolution. Are we better off than two years back? Yes. Are we there yet? I would say probably we are very early stages of maturity. Not about technology. It's applying the technology, adopting the technologies where the evolution is. Technology itself is moving on its own pace, and it will continue to move. Imagine we all are on social media. How much are we comfortable sharing on social media versus on office emails? I see there is a considerable amount of information that gets exchanged on social platforms than what it exchanges in the organization. That means there is some level of sensitivity around how people are able to operate, right? Lot of those things have to change. There are certainly behavioral dimensions. The workforce change dimensions which have to evolve as well. So how many people are comfortable to adapt into, yes, we want to move into this technology. It's not about an enterprise saying that, you got to adapt AI from tomorrow. People have to be comfortable as well. If they are not, then you've got to work through that change. So there's quite a few. Technology itself is also not there yet in all conditions. Consciousness of technology has to evolve.
Dylan Cosper:
So how can DPI help agencies adapt and evolve with changing policies?
Lax Gopisetty:
I mean, I don't think DPI itself will influence per se. But the way I would say is there is a learning which can happen from multiple countries. So today, Middle East is adopting DPI in a totally new way. You know, Saudi is building a new city, a new country on totally digital platforms. So for them, it's a greenfield evolution of new standards to be adopted. If I take some examples from Singapore has done as a smart country, where everything is very digital. So there are elements which we can all harness to see adoption of that regulations and guidelines. Second is it's an evolution of how we are going to, what we want to, what we want to enable in this interoperability. And infrastructure, as we build the digital public infrastructure, it will go through each swim lane will go through its own evolution and bring the people together because people will understand that they need to work together. But sometimes you have to nudge multiple times to make that happen. And DPI is that nudging construct forces people to operate more collaboratively, more interoperable. And hence the guidelines will also evolve to make sure that is feasible. I mean, as I said, two years back to now, it's already there. Two years back, there were no states which were creating that guidelines. If state is giving those guidelines, that means all the counties and cities are probably aligned to that. So there is already one cluster of cohort in a state is aligned to governing guidelines, and which means they are going to operate on one digital public infrastructure construct in that state, then they will all evolve into nationwide, or they will evolve into international. I think that's the way I see it, but it is only building up from bottoms up, or a ground level, and that actually makes it more meaningful, more adaptable, and more inclusive.
Dylan Cosper:
Well Lax, before we end, I'd love for you to pull out a crystal ball here and give me what's your vision for DPI over the next decade.
Lax Gopisetty:
I think DPI is the new future for helping the world to navigate the digital world in a better way, more responsible way, more ethical way. It will help technologies to interoperate. It will help humans to interoperate and collaborate more. It will also require humans to go through the learning curve, learning path of moving into the fully operating in the digital model. It will also help to bring new innovations in technology, which will align to responsible digital world, human-centric digital world in the future. I feel the next decade is said industrial era framework is going to be disrupted. I think it's already getting disrupted, but it will be more. And we all will… the world order will certainly get tweaked into a new structure, which we all will adapt in one way or other. But it is going to be much better than what we have seen is the way I would create a hope for. And we are all being part of the stakeholders in that.
Dylan Cosper:
And it's truly, I think that's a great point, is it's a unified outlook. Things are being tweaked, but it's being tweaked in the direction of being unified. Each and every one of us in this journey.
Lax Gopisetty:
Absolutely, and that's what makes it very exciting for me that we are being part of maybe the smallest drop in the ocean, that we are being part of that defining the future. And probably companies like Infosys and quite a few of our partners and our clients are all trying to... be in the leading edge of that space to drive that change and be part of that change. So that's exciting for me to be able to define the future and be part of it. Not just define it, but be involved.
Dylan Cosper:
Absolutely. Lax, thank you for joining and sharing your insights. It's clear that digital public infrastructure and agent AI are reshaping what's possible in public service.
Lax Gopisetty:
Thank you, Dylan. It's a very insightful conversation and look forward to the next one. Glad to be part of this journey.
Dylan Cosper:
Excellent, thank you so much for joining and I look forward to the next one as well. This has been another episode of the AI for Public Good series from the Infosys Knowledge Institute. Until next time, keep learning and keep sharing.