Kala Vairavamoorthy explores how Generative AI can empower water utilities in the Global South, using data to unlock new ways of working.
Across much of the Global South, water utilities sit on mountains of data – records of pipes and pumps, customer accounts and complaints, flow and pressure logs, project files and master plans. Decades of collection have produced archives that are both rich and silent. The irony is unmistakable: utilities are not short on data; they are short on understanding.
This is not a story of neglect, but of overload. Data were gathered for tenders, audits, and billing – useful at the time but never connected across systems. Over time, this created a quiet crisis of data blindness: the knowledge we already have cannot see itself.
Yet the professionals within these utilities have always known what needed to be done. They have the intellectual capacity, technical expertise and deep understanding of their systems to analyse and act on their data but rarely the human resources or time to do it. Each day’s operational pressures demand immediate attention: pumps to fix, leaks to trace, customers to serve. The deeper analytical work – the kind that reveals patterns and long-term solutions – sits patiently in the corner of the control room, waiting for a quieter day that never comes. These were the important but not urgent tasks – acknowledged, but endlessly deferred by the pressures of the urgent.
Today, a new technological chapter is opening that can change this reality. Generative Artificial Intelligence (GenAI) and the large language models that power it are finally lowering the barriers that kept data locked away. For the first time, human language has become the programming language.
And importantly, utilities in the Global South already have the know-how and contextual insight to embrace this innovation. With even modest data foundations, GenAI can empower them rapidly – turning scattered information into collective intelligence and enabling more precise, confident and timely decision-making.
This is not just another digital upgrade; it could be a genuine game-changer for utilities across the Global South, offering a leap in capability rather than a slow climb.
Collecting without connecting
Across utilities, the same story repeats. Pipe drawings sit in one office; customer records in another. Leak reports are handwritten. Complaint registers live in Excel sheets that no-one has opened in years. Master plans, often developed with donor support, compile enormous amounts of data on existing systems – painstakingly gathered and documented – even though the expansions they envision are rarely implemented. These are pieces of a puzzle that was never assembled.
Meanwhile, data collection has always had clear motives: to demonstrate transparency in tendering, ensure accountability for spending, and meet design standards and regulatory expectations. Added to this, utilities have become highly sophisticated and systematic in collecting and analysing data for billing and revenue purposes, deploying aggressive efforts to ensure customers pay for what they consume.
If only the same level of discipline and innovation were applied to understanding the performance of their water systems and the quality of service delivered to customers, the results could be transformative. The know-how clearly exists – it’s simply not directed toward this purpose. The result is a landscape of uncoordinated fragments.
It’s like a vast library where every department keeps its own locked shelf – a goldmine of insight where most of the books are never opened. The next leap is not more collection – it is connection.
The cost of data blindness
This fragmentation comes at a high price. When information systems don’t speak to each other, utilities lose visibility and that invisibility costs money every single day.
Huge sums can be spent replacing pipes that still have years of useful life. Work orders are issued for repairs that never needed to happen. Technicians waste hours searching for the history of a pump or valve, only to discover that the one person who knew its details retired years ago.
These are not failures of technology but of knowledge management. Utilities end up drowning in information but starving for insight.
The 15% we see – and the 85% we ignore
Utilities typically use only a small slice of their total information – perhaps 15%. It forms the visible tip of a much larger iceberg. The other 85% is a graveyard of missed opportunities – not useless, just untapped.
Dashboards can signal a drop in pressure but not explain why it happens every July. They can show a revenue dip but not the structural reason behind chronic non-revenue water. The real intelligence – the unseen 85% that holds context, causality, and meaning – lies beneath those dashboards, unstructured, messy and incredibly valuable. Until now, analysing this mountain of text, images, and notes required armies of data clerks and programmers – a task far beyond the human and financial capacity of most utilities in the Global South. GenAI changes that balance entirely. It can read across formats, languages, and styles, connecting what people wrote, saw, or said into coherent patterns.
Crucially, this does not require a perfect database. Utilities across the Global South can start where they are, with what they already have. They don’t need to spend years cleaning every record before extracting meaning. GenAI tools are designed to handle imperfection – to interpret, summarise, and correlate.
This is an opportunity to accelerate straight from data collection to insight, without the slow crawl of conventional digitisation.
GenAI – the great equaliser
Unlike earlier waves of automation that required coders and consultants, GenAI lowers the barrier to entry. Anyone can query the system using natural language.
So, an operations engineer might ask:
- “Which neighbourhoods have the most leak complaints?”
- “Summarise the past five years of pipe bursts by material and contractor.”
- “Which pump stations show recurring downtime during peak demand periods?”
The answers return as readable text, charts and summaries – turning the data landscape into a living conversation. Insight is no longer the privilege of those who can code; it belongs to those who know the system.
The same capability extends beyond analysis to practical, everyday operations:
- Leak and non-revenue water management: by combining pressure data, leak reports, and customer complaints, GenAI can reveal hidden losses and pinpoint high-risk zones.
- Predictive maintenance: analysing past failures enables the AI to forecast where pumps, valves, or pipes are likely to fail next – when staffing is thin, prediction becomes manpower.
- Alarm triage and workflow efficiency: the system can sift through floods of meter or SCADA data and alerts, distinguishing between noise and genuine issues, reducing operator fatigue.
- Work-order validation: duplicate or unnecessary maintenance requests can be detected automatically, saving both time and money.
- Automated Reporting: AI can draft technical summaries, regulatory reports or project updates directly from raw logs, freeing staff to focus on solving problems rather than describing them.
A field engineer, a billing clerk or a district supervisor – all can become analysts in their own right.
For utilities across the Global South, this accessibility is revolutionary. Low digital maturity is no longer a barrier. Small utilities can run open-source, lightweight AI assistants on local servers or even smartphones. The most resource-constrained organisations can begin experimenting with their own data in months, not years.
The human factor – capturing lost wisdom
An important part of the GenAI opportunity is understanding that not all knowledge lives in files. Some of the most valuable insights are tacit knowledge – the unwritten experience carried in the minds of operators, fitters and inspectors who know their systems by feel.
Think of a library: the books hold facts, but it’s the librarian who knows where the real gems are and which ones matter most. In the same way, every utility has its librarians – the veteran pipe fitter who can sense a leak before a sensor does, or the billing clerk who notices irregular consumption patterns weeks before the system raises an alert. These instincts are data too, formed through repetition, mentorship and intuition.
Yet as staff retire or move on, that living memory evaporates. Every departing worker takes a library with them. In a sector where institutional memory is one of our greatest assets, losing tacit knowledge is like losing the map to our own system.
GenAI offers a way to preserve and amplify this human intelligence. By recording short interviews, field debriefs, or even smartphone videos, utilities can feed these experiences into AI systems that store and cross-reference them with operational data. Tomorrow’s AI won’t just read text – it will listen to voice notes, interpret photos and correlate them with field events, creating a searchable memory of human experience.
This has profound implications for day-to-day operations:
- Predictive maintenance improves when human intuition – what engineers suspect – is integrated with historical failure data.
- Workflow efficiency increases when the system learns from how experts triage alarms or handle complex repairs.
- Training and onboarding accelerate when new staff can learn directly from the preserved tacit knowledge of their predecessors.
Building a culture of shared knowledge
To truly capitalise on this opportunity, there needs to be an organisational shift in mindset. Technology alone won’t create intelligence; it needs a culture that values learning over paperwork. Every repair, inspection and conversation should be seen as knowledge to be captured and reused.
Encouraging senior staff to ‘mentor’ AI systems – by recording their methods, stories, and decisions – can bridge generations of experience. Rather than fearing that machines will replace them, staff can recognise that these tools will remember them.
In this outlook, AI does not replace expertise – it preserves it. Over time, tacit knowledge once locked in individual heads becomes a shared asset that strengthens the utility’s resilience. This is where the partnership between humans and machines truly begins.
And this is where GenAI becomes a true game-changer for our water utilities: imagine feeding all those dusty logs, contractor records and the everyday wisdom of our operators into one intelligent system – no complex programming required, just simple, natural language.
In the near future, AI will become a quiet colleague – one that remembers, summarises and alerts, but doesn’t tire or forget. For water utilities operating with limited resources, this partnership could mean the difference between chronic crisis and steady control.
The goal is not automation for its own sake, but augmentation: giving people the ability to see their system whole, to anticipate instead of reacting, to learn continuously from their own experience.
From insight to action
As utilities grow more comfortable using GenAI to uncover patterns in their data, a natural question follows: can these systems also support decision-making?
Imagine a small utility facing chronic water loss. The manager knows from local observation that leaks are common and pressure is erratic. In the past, the only option would have been to hire external consultants – an expensive process that often begins from scratch, repeating analyses already captured in reports and records.
Now, that same manager could describe the situation to a GenAI assistant: outline the system layout, share field notes, and provide anecdotal evidence of leaks or shortages. The AI could then help structure a plan – suggesting how to verify those observations, which data to review first and what preparatory steps local staff could complete before consultants are brought in. By identifying low-cost, high-value groundwork, utilities could save both time and money while making external expertise more focused and effective.
Used prudently, this kind of interaction transforms GenAI from a passive information tool into an active thinking partner. It doesn’t replace professional judgement – it refines it. The real power lies in helping managers frame better questions, test their assumptions and approach experts with clearer priorities and stronger evidence in hand.
But this is also where promise and caution must meet. When GenAI begins to move beyond the data supplied by the utility – venturing into prescriptive advice – it risks blending inference with fact. Confident wording can conceal uncertain reasoning. For that reason, GenAI should be used within clear boundaries, relying on verifiable data and transparent logic. When used this way, it becomes not a source of prescription, but a partner in structured thinking – helping managers frame better questions long before the experts arrive.
The lesson is balance. GenAI can help under-resourced utilities think more strategically, prepare more efficiently and engage consultants more intelligently. But it should guide decisions, not make them. In this way, utilities can benefit from AI’s analytical reach while keeping human expertise – and accountability – at the heart of every choice.
Call to action
This decision-support role is the next frontier, where utilities begin asking GenAI not only what the data say, but what to do next. That promise must be approached with humility. Hallucination, bias, or overconfident advice can mislead even well-intentioned users. The responsible path is to pair GenAI’s reasoning power with human judgement – to prepare, question, and verify, never to replace expert review.
I have been fortunate, during my academic career, to have many students who have gone on to become utility leaders. Across the utilities I have worked with, I have seen how quickly people adapt when given the right tools. Their curiosity, courage, and willingness to embrace innovation have convinced me of the great potential that exists across our sector. With the right support and mindset, even small teams can unlock entirely new ways of working.
GenAI, if used wisely, could become one of the most empowering tools our sector has ever known – not because it thinks for us, but because it helps us think more clearly.
The divide in the water sector is no longer between rich and poor utilities, but between those who use their knowledge and those who don’t. For the Global South, this is a rare moment to move ahead – not by copying the digital pathways of others, but by designing approaches that are lighter, faster and better suited to local realities.
As I have said elsewhere, we live in an era of infrastructure in flux – a time when transformation goes far beyond new pipes or new plants. It’s about how we use every tool we have – physical, digital, and human – to build a resilient, intelligent and circular water future. GenAI will be a critical part of that journey.
IWA – Turning data brilliance into action
The International Water Association can help make this journey a shared one. As a global convenor, IWA transforms innovation into collective progress, connecting utilities, researchers, and practitioners who are learning how to apply GenAI responsibly and for the public good.
This is where IWA can play its strongest role and where it already excels. It can help utilities navigate this new landscape by curating and critically reviewing GenAI platforms, assessing their suitability for different contexts – recognising, as ever, that there are different horses for different courses.
Beyond convening, IWA can add tangible value by developing guidance for effective and inclusive AI use: from frameworks for responsible data preparation and interrogation, to libraries of tested prompts and workflows, including multilingual examples that make GenAI accessible across linguistic and cultural contexts.
IWA can also nurture a global community of practice, where utilities exchange lessons, exemplars share their journeys and peers learn from both successes and setbacks. Through its webinars, publications and conferences, IWA will continue to facilitate open dialogue and shared understanding – helping utilities move from curiosity to confidence, and from isolated innovation to collective intelligence.
The future of the water sector will not be defined by how much data it holds, but by how intelligently it uses it. The tools to listen, interpret, and act are already within reach. The future will not wait for perfect systems; it belongs to those who start – those who transform dormant data into shared wisdom.
And it begins now – with the courage to ask better questions of the knowledge we already possess.
The author
Kala Vairavamoorthy is CEO of the International Water Association
