top of page

Portfolio Highlight - Our Investment in Voda.AI

Updated: May 17, 2023

By Mark Legare, Senior Associate at Innospark Ventures

Certain industries match well with certain founder profiles. A statement as equally true as it is banal. Within the vast domain of B2B enterprise, “breaking in” carries remarkably different motions depending on who it is you are targeting. Often, there is a strong correlation between how niche an industry is and how cumbersome its sales cycle portends to be. Specialized knowledge born from prior experience, or deep domain expertise of some type, greatly enables effective navigation of these more difficult adoption hurdles. One way to decipher the relative opaqueness of an industry, and the ensuing sales cycle founders face, is to gauge the extent to which an industry has its own language, processes, and rules. A clear example: selling to public and private water utilities. Large, niche, and relatively obscure, the ~$90Bn North American water infrastructure technology market is ripe for disruption but also has several trapdoors along the way. As we dug in to find AI-driven opportunities in the space, we were quickly reminded of just how beneficial it is to have a domain-driven CEO at the helm. We found that in George Demosthenous of

“I come from water”. At first, we did not know if George was offering a metaphysical trope or a background statement. It could have been either. After nearly a decade in the water industry, including three years as the head of software products for Mueller Systems, George co-founded to offer a purely software-driven approach to alleviate non-revenue water loss (NRW) for utilities. Each year, the United States loses roughly ~30 gallons per capita per day. The math on that loss is staggering. With water priced at roughly $.01/gallon, estimates have total NRW loss anywhere from $7Bn to $30Bn annually. Losses stem from a simple reason: pipes break. In fact, each year there are ~250,000 breaks across the ~2 million miles of pipeline, with the US losing approximately 25% of all treated drinking water. To compound matters:

1. US pipes will begin to “come due” over the next five to 10 years, as the American Water Association predicts ~30% of all pipes will need to be replaced by 2040.

2. Water itself is becoming increasingly scarce (particularly in the Southwest) amidst a rise in climate-related events.

3. On average, each pipe break costs $3,000 in labor and materials to fix, and with over 250,000 breaks/year, that’s $750M the US is spending just to fix breaks and leaks.

4. The U.S. Bureau of Labor Statistics estimates a ~7,000 water utility worker shortfall over the next decade.

With the cost of water rising 43% over the last decade, faster than any other household utility, it’s not surprising that many experts are warning of an impending “water crisis”.

Though the problem of water loss may seem simple, solutions are scarce and complex. There are two main drivers behind this: (1) capital allocation, predicting leaks and prioritizing pipes for replacement is a surprisingly difficult task and (2) changing century-old processes and moving from reactive to proactive requires interfacing with many different groups who share vastly different incentives (e.g., utility engineers, planners and operators, state & local governments, and the Department of Transportation). We believe success in both areas is a prerequisite for building a real business in the space.

Given the myriad of challenges,’s AI-driven risk modeling is a gamechanger. Traditional methods of predicting which pipes are most vulnerable to breaks are expensive and inefficient. For example, “smart” devices are invasive, costly, less accurate in PVC pipes (22% of the entire network), and prone to getting stuck and/or lost. Aerial data has also been attempted but has yet to prove efficacy. Other more traditional methods include having utility workers walk along the pipe with sensors or having workers physically cut into a pipe, pump traceable gas into it, and then run along the pipe to see if colored smoke appears. Unlike these methods, is pioneering a new form of pipe surveillance that is 100% noninvasive, affordable, and highly accurate.

Using sophisticated AI and ML,’s patented approach examines a wide range of variables, such as pipe age, type, location, and local soil conditions, to predict when and where a pipe will burst or whether a pipe has a high likelihood of being made with hazardous lead material. Before making each prediction, the company’s algorithms process large amounts of data through a rigorous ML-driven QA algorithm, which cleans and harmonizes disparate data sets. This initial step is crucial, as utility data varies greatly between different providers. Next, the results are sent to a hyperparameter model that operates uniquely for each customer.’s proprietary method selects the best algorithm or combination of algorithms for every client, yielding the most accurate results in the utility industry. Given the breadth of technical expertise, customers large and small are adopting and seeing immediate value. Using’s technology, private and public water utilities, consulting and engineering firms, and governments and agencies can now find where lead pipes remain, launch a complete virtual assessment, and run AI-driven capital management projects.

The recent passage of the “Infrastructure Investment and Jobs Act”, which includes $50 billion in federal grant money for water infrastructure projects, makes it clear that for a solution like this, the time is now. We tried mapping out what these differing incentives look like but found that in our discussions with various buyer groups, it was a challenge for us to pin down the exact how, when, and why buyers in this market engage with certain vendors over others. As the company has begun to rapidly expand (now serving 15 states and 6 countries), the reason (& lesson) comes back into focus — not only does have exceptional technology, but just as important, the founders serve as a Rosetta Stone to different actors across these various segments.

At Innospark, we get excited about the idea of ‘doing good and doing good’ — George and team are not only building an exciting business, but beyond that, they are living the mission of “AI for Good”. As we continue to wade into unpredictable waters given the rapidly evolving AI landscape, we are compelled to take our roles (but not ourselves) seriously when it comes to ushering in the next wave of computational intelligence. Along with basic questions like product-market fit, market potential, defensibility, and strength of team, we also ask ourselves: “what happens if this technology succeeds?” One thing we are sure of: the world is a better place with

If interested in connecting with the team, please contact them here.

Innospark invests at the intersection of AI and major industries such as Healthcare, Life Sciences, and B2B/Enterprise. If you’re an innovator in that space, reach out to us here.


bottom of page