Download the full interview here. 

“Forward-looking utilities understand that modernization cannot come at the expense of the communities they serve,” said Deepak Garg, Chairman, Founder, and Group CEO at SEW.AI.

The vertical AI company supports energy and water companies with a cloud-native foundation, providing context-specific solutions to support the unique grid needs that utilities rely on. Garg leads the creation of industry-specific agentic AI platforms, aiming to both efficiently assist utilities providers and the demands they face, while advocating for global sustainability.

In a conversation with ASUG, Garg explained how the SEW.AI Platform works with SAP to drive further data-informed insights. He also addressed the numerous pressures of the utilities industry—and how AI can help better support the communities these companies serve.

This interview has been edited and condensed for length and clarity.

Q: Many companies run SAP as the backbone for ERP, billing, and asset management. How does SEW’s platform sit alongside that infrastructure, and what does the integration look like?

Our partnership with SAP spans decades; it’s a partnership we are proud of and one that has been built on a deep alignment in how we view this industry. SAP has consistently been the backbone for energy and utilities, bringing structure, reliability, and scale to some of the most complex operations in the world. As the industry has evolved, that foundation has only become more valuable. At the same time, what has changed is the environment around it. Utilities today are operating in a far more dynamic, data-rich, and unpredictable landscape. The opportunity we saw was not to rethink that foundation, but to extend its value.

SEW.AI Platform works alongside as extended SAP platform, and is deeply integrated into its data and processes, bringing additional layers of intelligence across the energy and utility value chain to enable faster, more informed decisions with greater context and speed.

It is the ecosystem model we are creating—connecting customers, workforce, businesses, and the grid—and our clients are already living that reality. Leading utilities such as San Diego Gas & Electric (SDG&E) and many more global energy and utilities providers are a powerful joint success example and story: through our integrated platform strategic partnership, they are driving continuous innovation across digital experiences and service modernization. Similarly, at Dominion Energy, this vision is reflected in deep focus on redesigning the experience around multi-persona and tailored journeys for residential customers, enterprise users, landlords, and agencies.

The value of SEW.AI and SAP platform is that it multiplies ROI. Our clients don’t have to choose between their enterprise backbone and AI-driven vertical intelligence. The two deeply vertical-focused platforms work together, and that is precisely why it is driving this industry transformation at this scale. In a world where utilities are under pressure to modernize faster than ever, driven by data center demand, electrification, and grid instability, having these two tech stacks working in concert is not a luxury—it is a survival imperative.

Q: You recently introduced SEW.AI COSMOS Vertical AI Native Platform. What sparked this vision, and how has it taken shape over the past year?

With what we are witnessing now across the industry and in ongoing conversations with energy and utility leaders, partners, analysts, and our teams across regions, it became increasingly clear that the industry is at a defining inflection point. Energy and water are no longer background services; they sit at the center of economic growth, resilience, and community well-being. At the same time, rising demand, the rapid growth of data centers, the acceleration of AI, and mounting pressure on infrastructure are fundamentally reshaping expectations from utilities. It was in this context that the vision for SEW.AI COSMOS – Vertical AI Native Platform took shape, not as a response to trends, but as a purposeful step toward enabling a more connected, intelligent, and resilient utility ecosystem.

When we first spoke about SEW.AI COSMOS – Vertical AI Native Platform at the WE3 Summit, the focus was on what vertical AI could mean for this industry as an operating model. Since then, that vision has steadily matured through real-world platform use cases across customer, workforce, and grid transformation initiatives. Today, SEW.AI COSMOS stands as a unified, vertical AI native platform that brings together customers, workers, assets, operations, and grid intelligence into a continuously learning ecosystem, purpose-built for the complexity of energy and utilities.

More than the technology itself, it reflects a long-term commitment to innovate with intent and to partner closely with utilities as they navigate this transformation.

Connect with Deepak Garg on LinkedIn. 

Q: SEW.AI COSMOS is built around what SEW.AI calls agentic vertical AI, which implies the AI is executing, not just recommending. Where in the utility workflow are AI agents currently taking actions without human approval, and where have you deliberately kept a human in the loop?

Traditional automation was built around tasks. Agentic vertical AI is built around outcomes. That distinction matters enormously in energy and utilities because utility operations are deeply interconnected. A decision made in customer operations can impact workforce dispatch. A grid event can impact billing, outage management, and safety operations simultaneously. Intelligence cannot remain siloed.

What we are building with SEW.AI COSMOS is a platform where vertical AI can coordinate signals across the enterprise, identify patterns humans cannot process fast enough, and automate lower-risk operational decisions in real time. But we are equally deliberate about where human judgment remains essential.

In utilities, there are decisions that carry operational, regulatory, and public safety consequences. Those should never become black-box automation exercises. That is why our philosophy has always been people plus AI, not people versus AI. AI should elevate human expertise, not remove accountability from the system.

In practice, vertical AI can autonomously optimize workflows, prioritize operational actions, surface risk intelligence, and coordinate system responses across interconnected operations. But when decisions impact safety, critical infrastructure, or customer trust, human oversight remains essential. The future of utilities will not be fully autonomous systems. It will be intelligent systems where human judgment and machine intelligence continuously strengthen one another.

Q: In a world increasingly driven by horizontal AI platforms and general-purpose intelligence models, why is vertical AI becoming essential for the energy and utilities industry?

Utilities do not operate in generalized environments. There is a fundamental difference between horizontal AI capability and vertical AI intelligence, and that difference is the entire reason SEW.AI exists. A model that has been trained on utility-specific data, that does understand the physics of power distribution, the regulatory constructs of rate cases, the operational reality of managing aging infrastructure in climate-stressed environments—that model has the depth required to drive real outcomes.

Horizontal AI cannot give you the domain intelligence that comes from years of working inside this industry, building models against real utility data, and understanding what a “five-star” customer experience looks like operationally and financially for a regulated energy or water company.

With SEW.AI COSMOS – Vertical AI Native Platform, we are building the intelligence layer that translates raw enterprise data and AI capability into decisions and actions that are meaningful in the context of operating a grid, managing a watershed, or serving a datacenter. Our role is not to compete with generalized platforms—it is to unlock the value of both existing and new systems for an industry that has some of the most complex operational requirements on Earth. A utility running a grid that must balance supply and demand in real time, respond to storms, meet regulation mandates, and keep bills affordable for customers who are increasingly financially stressed. That utility needs a partner who understands all of that complexity deeply, not a general-purpose tool that needs to be configured from scratch by people who have never worked in the sector.

Q: You’ve noted that 34% of North Americans qualify as low-income, and that bills could rise significantly as AI-driven demand hits the grid. Can AI itself help keep modernization from becoming a cost that gets passed through to the customers least able to absorb it?

This may become one of the defining questions of the energy transition. As infrastructure demand accelerates, the traditional response has been straightforward: build more infrastructure and recover the cost through ratepayers. But the current scale of modernization raises a much bigger question around affordability and equity. AI can help change that equation, but only if it is applied intentionally.

We are delivering capabilities that proactively identify at-risk customers before they reach a crisis point: customers who are likely to disconnect, who qualify for assistance programs they haven’t enrolled in, or who could benefit from a targeted efficiency intervention that lowers their consumption and their bill. This is vertical AI being applied not just to operational efficiency but to social outcomes. For example, SAP and SEW.AI are proud to work with utilities like DTE Energy, which is leveraging technology to strengthen engagement with vulnerable customers through financial assistance and energy affordability initiatives. This includes more proactive outreach, improved pledge management, and targeted support to help customers maintain continuous utility service.

Utilities operate with public trust. The forward-looking utilities understand that modernization cannot come at the expense of the communities they serve. Intelligence should not only make utilities more efficient. It should help make them more equitable.

Q: You’ve described SEW.AI COSMOS – Vertical AI Native Platform as taking the hard path of building for 50 years rather than five quarters. What lessons have you learned by taking the hard path?

The first lesson is that depth compounds. When you build with genuine depth—in domain knowledge, in data architecture, in regulatory understanding, in customer relationships—you create capabilities that are very difficult to replicate quickly. We made a deliberate choice early on not to build the product that was easiest to demo but hardest to deploy at scale. We built for the operational reality of a utility: complex data environments, legacy systems that predate the internet, workforces that may be skeptical of technology that doesn’t understand their daily reality, and regulators who need confidence that AI recommendations are explainable and auditable.

That was hard.

The second lesson is that trust is the only durable currency in this industry. Utilities operate critical infrastructure that billions of people depend on for their safety and quality of life. They do not give their trust easily, and they should not. Every time we have earned that trust, by being honest when something didn’t work the way we projected, by staying through implementation challenges rather than moving on to the next sale, by building relationships with operators and field crews and not just executives, it has opened more doors.

The third lesson, and perhaps the most important, is that the hard path attracts the right people. The people who join SEW.AI are not here because it was the easiest option. They are here because they believe in what we are building and why. That is what I call the “Super Warrior Mindset”—a mindset rooted in ownership, resilience, and the conviction to solve the hardest problems facing the industry, even when the answers do not exist yet. It is about building with long-term responsibility, continuing to innovate when the path is uncertain, and understanding that the work matters because millions of people ultimately depend on the systems that we help power and modernize. That mindset has shaped everything we have built so far, and I believe it will define everything we build in the decades ahead. Building for 50 years means you are always committed to something larger than the next quarter.

Q: When a utility CIO is evaluating AI partners, what are the red flags that a solution is generic tooling repackaged for the industry rather than something built with real depth?

The stakes of getting this wrong are not abstract. A poorly chosen AI partner can consume years of implementation effort, erode internal confidence in the technology, and leave a utility further behind than when it started. So, when I advise CIOs on how to evaluate what they are being shown, I tell them to ask hard questions in four specific areas:

• The pitch leads with technology, not outcomes. If an AI provider’s primary reference is their model architecture, their foundational LLM, or their cloud infrastructure rather than specific utility problems solved in production with measurable results, that is your first signal. Ask directly: where have you been running in production for more than two years, and what did it cost the utility before you arrived versus after? If they cannot answer that precisely, the depth is not there.
• Ask to speak with operations, not just IT. Operational staff know within minutes whether a technology actually understands their work. A provider who steers you only toward IT leadership during reference calls is telling you something. The people closest to the work are the most honest evaluators of whether an AI solution was built for the industry.
• Their data strategy is not domain-specific. Ask them to explain what a utility’s data model looks like. How do they handle the integration between AMI, CIS, SCADA, and work management systems? If the answer involves building custom connectors from scratch for each client, you are looking at a horizontal tool trying to fit a vertical problem. A genuinely vertical platform arrives with that integration architecture already solved.
• The AI cannot explain its own reasoning. Explainability in this industry is not a nice-to-have—it is a regulatory and operational necessity. If an AI recommends deferring maintenance on a critical asset, the operator must understand why and must be able to challenge that reasoning with their own knowledge. Ask the vendor directly: can your system explain, in plain language, why it made a specific recommendation? Can that output be audited? Can a regulator review the basis for a decision that affected rates or reliability? If the answer is vague, walk away. Black-box AI has no place in infrastructure management.

Q: The workforce conversation in utilities usually centers on giving field crews better tools. But institutional knowledge is retiring out of the industry at scale. Can AI realistically capture what a veteran lineworker knows, or is that a different kind of problem?

This is precisely where our people-plus-AI philosophy becomes most tangible. Field operations are where utility decisions become physical, immediate, and safety-critical, and the knowledge walking out the door with retiring veterans is not just experienced, it is institutional infrastructure. AI cannot fully replace what a 30-year lineworker carries in their instincts. But it can capture patterns from decades of work-order histories, asset behavior, and incident data, and put that intelligence directly in the hands of the next generation of field crews in real time. The goal is not replacement. It is amplification. 

In practice, this shows up across the entire field lifecycle:

• Work is prioritized not just by urgency, but by combining risk, skill fit, and environmental conditions.
• Crew assignment and routing move beyond logistics into capability matching, ensuring the complexity of the task aligns with the experience in the field.
• Before a truck even rolls out, pre-job risk intelligence surfaces potential hazards and operational constraints.
• During execution, field teams are supported with real-time, AI-driven guidance that flags safety conditions and procedural steps, while still requiring human confirmation for every critical action.
• And as conditions evolve, there is continuous coordination between field crews and control centers, ensuring decisions are made with full situational awareness.

One of the largest utilities in North America and a long-standing client, Pacific Gas & Electric (PG&E), has a field workforce operating across a service territory of extraordinary scale and complexity. Their workforce is connected to our unified workforce experience AI platform documentation, operational guidance, and compliance context directly into the field environment, providing on-the-job training. Not digitizing knowledge for reference, but operationalizing judgment at the point of action, so that institutional intelligence does not retire when people do, but continues to strengthen how the industry is built, maintained, and secured. And this is what people plus AI looks like in action.

Q: SEW.AI operates across dozens of countries. Are there markets where utilities are further ahead on digital maturity than their American counterparts, and what should U.S. utilities be learning from them?

Working alongside 470-plus utilities across 47-plus countries gives us a vantage point that genuinely shapes how we build. Every region brings its own objectives, its own opportunities, and its own possibilities. And behind every utility is a community of people—billions of people—who deserve to be engaged, empowered, and educated by energy and water that runs their daily lives. That breadth of perspective shapes how we think, how we build, and what we bring back to every country we operate in. Regulatory models, customer expectations, and infrastructure realities in different countries dictate a pace of innovation that creates lessons worth carrying back.

For instance, Sydney Water, the largest water provider in Australia, has set a standard for how a utility can reimagine the customer relationship around conservation and digital engagement, treating water not just as a service but as a shared responsibility.

In India, Tata Power, one of the most iconic brands in the country, embarked on a bold digital transformation journey with SEW.AI, launching an AI-powered customer experience platform across its multiple state distribution companies and becoming the only utility in India to launch a Super App that unifies services across all its group companies into one seamless, digital-first experience.

In Latin America, Gasmig in Brazil is redefining what customer experience means for the region, deploying our digital platform including the vertical AI agent built specifically for energy and utility use cases, from billing and payments to service requests and digital account management. These are not pilot programs. These are utilities that made a decision to lead, and they are raising the bar for what the industry is capable of delivering everywhere.

Visit the SEW.AI website. 

Q: Water shows up consistently in SEW.AI’s mission but gets far less airtime than energy in digital transformation conversations. Is the water industry meaningfully further behind on digital maturity?

Water and energy have always been equal parts of our mission. It is a conviction we have held since the beginning. The conversation around water in digital transformation circles is quieter than energy, but I would not frame it as simply being behind. The water industry is navigating a different set of pressures: aging infrastructure, water scarcity, tightening regulatory standards, and communities that are deeply connected to water as a public good in a way that is distinct from how they relate to energy. What is changing, and changing fast, is the recognition that digital transformation in water is no longer optional. Climate stress, population growth, and rising expectations from residents are forcing the issue in ways that are accelerating the conversation meaningfully.

Oklahoma City Water Utilities Trust (OCWUT) is a powerful example of a water utility that moved with real clarity and purpose. Unifying water, wastewater, and waste into a single resident experience, one that is multilingual, secure, and seamless, meant dismantling years of fragmented systems and rebuilding around the person being served. This work was recently recognized with an industry award, and while awards are not why we build, this one matters because it signals something larger: that the water sector is beginning to set its own bar for what resident-first digital innovation looks like. OCWUT did not wait for the energy industry to show the way. They defined their own standard, and that is exactly the kind of leadership that moves an entire sector forward.

Q: When you look at the next three to five years—with data-center demand, electrification, and climate volatility converging on the grid—what would you tell ASUG members still deliberating about where to start?

What makes this moment so extraordinary for the utilities industry is that, for the first time in decades, energy and water have moved to the center of nearly every major transformation happening in the world. The rise of AI, the acceleration of electrification, the expansion of data centers, the growth of distributed energy—none of it advances unless the grid evolves alongside it. Utilities are no longer operating quietly in the background of progress; they are becoming the infrastructure enabling the next era of human and economic advancement. That is what makes this such a defining moment for the industry, and honestly, such a meaningful time to be building alongside utility leaders who carry the responsibility of long-term reliability every single day.

This is also why partnerships like SAP matter so deeply in this transformation. Utilities already possess decades of operational intelligence inside these systems: asset histories, customer records, workforce patterns, outage data, operational workflows. That foundation is incredibly valuable because it makes transformation practical, not theoretical. Utilities do not need to rip apart the systems they trust to run their business. They need an intelligence layer that can connect to what already exists, learn from it, and turn it into operational decisions that teams can act on in real time.

The future will not belong to isolated systems or disconnected AI deployments. It will belong to utilities that build connected intelligence ecosystems, where customer operations, workforce intelligence, grid assets, and business systems continuously learn from one another. That is the shift we have been focused on for years: helping utilities move from fragmented operations to a model where intelligence flows across the entire value chain. When that happens, AI becomes far more than automation. It becomes the ability to anticipate demand, optimize energy movement, strengthen resilience, and operate the grid with a level of precision and coordination that was previously impossible.