In the bustling corridors of Veolia, the buzzword isn’t just sustainable water management anymore. It’s AI. Meet Andrew Collier and his digital team, who are not just transforming water treatment but setting audacious goals to marry artificial intelligence and LLMs with Veolia’s vast network of industrial solutions.
Their ambition? To turn tens of thousands of cooling towers (and countless more assets) into data goldmines and push the envelope on how the world perceives water technology. Let’s explore it:
with 🎙️ Andrew Collier – Digital Solutions Executive at the Veolia Water Tech Zone
Resources:
🔗 Come say hi to Andrew on LinkedIn
🔗 Veolia’s water tech zone website
🔗 Introducing Hubgrade on this microphone… 4 years ago!
🔗 Already praising Veolia’s agility (with Glenn Vicevic) last year
is on Linkedin ➡️
Full Video:
Table of contents
Behind the Digital Curtain
Andrew Collier, Digital Solutions Executive at Veolia’s WaterTech Zone, is at the forefront of this grand experiment. He recalls the rapid transition from conventional processes to embracing AI and large language models (LLMs). “AI won’t steal your job,” Collier muses. “But someone who knows how to use AI will.”
This statement is a testament to the evolution and integration of new technologies within Veolia—a company that prides itself on innovation and forward-thinking strategies.
Marrying Tradition with Cutting-Edge Solutions
Veolia’s foray into AI isn’t a half-hearted stab at modernization. They are developing secure, internal versions of AI models, crafting AI policies, and collaborating within their corporate structure to tailor these technologies to their intricate needs.
Collier’s team is a paradigm of agility to that extent, working in sprints to iterate on software tools every two weeks, ensuring that the end products are both user-centric and highly functional.
AI in Action: From Internal Tools to Marketable Solutions
One of the standout products mentioned by Collier is the Hubgrade Water Footprint. This tool aims to provide a comprehensive view of a plant’s sustainability impact, a crucial feature in today’s ESG-centered world. Companies can now visualize, monitor, and optimize their water and energy footprints in real time, using actionable data. “This tool aligns ESG leaders with their sustainability targets effectively,” Collier explains.
The digital team’s strategy extends beyond mere innovation for innovation’s sake. They meticulously validate AI models through curated data sets and expert feedback sessions, ensuring the outputs are reliable and free from the infamous ‘hallucinations’ that AI can sometimes generate. This isn’t just about keeping up appearances; it’s about achieving tangible, verifiable results.
The Broad Brushstrokes of Transformation
Collier elucidates the complex yet efficient structure within Veolia. The company is not only an industry giant but also a nimble innovator. By dissecting narratives like the integration of cooling tower data, Collier demonstrates how Veolia leverages its existing assets to pilot advanced solutions. With over 10,000 cooling towers plugged into their system, Veolia has set a new standard in data utilization, enabling comprehensive analytics that can detect anomalies, forecast issues, and provide empirical benchmarks.
This drive for operational efficiency exemplifies what Collier calls ‘proof, not promises.’ The numbers speak for themselves: in the last 12 months, Veolia’s projects have resulted in significant savings—167 million in operational costs, 240 million kilowatt hours of energy, and 14 billion gallons of water.
From Policy to Practicality
Veolia’s reach into AI hasn’t been without its hurdles. The all-encompassing reach of Collier’s digital arm necessitates a delicate balance of internal policies and external expectations. While internal experimentation is a given, extending these capabilities to external customers requires not just technical robustness but also legal finesse.
The potential for a fully automated plant is tantalizing but still on the horizon. For now, AI serves as an enhancer rather than a replacement. “Think of it as pairing expertise with efficiency,” Collier says, painting a picture where human intellect augments rather than sidesteps technological processes.
Foresight and Flexibility: The Future of Water Management
Where does this leave the future of water technology? Collier and his team are already thinking beyond the obvious. Generative design powered by AI sits prominently in their future pipeline—a tool that could revolutionize how water treatment facilities are designed by offering various optimized scenarios based on countless parameters.
The ultimate goal, however, remains constant: to develop integrative systems where data informs design, operation, and eventual optimization, creating a cyclic and symbiotic relationship. Veolia’s transition into AI isn’t merely for show; it’s set to provide comprehensive solutions with profound implications on the global water management stage.
From Internal Adoption to Industry Standards
Interestingly, Veolia’s advancements also signal a broader shift in the industry. Smaller startups and corporates alike stand on the brink of integrating AI at a level previously thought unviable or too complex. Collier notes that, while daunting, the potential for collaborative innovation across sectors could pave the way for even more rapid advancements.
Proof, Not Promises
In essence, Veolia’s journey under Andrew Collier’s digital stewardship illustrates a compelling narrative of growth, adaptation, and unwavering commitment to advancing water technology. It’s a story of one company not just keeping pace with technological advancements, but setting benchmarks for what is possible when environmental stewardship meets digital innovation.
As Veolia’s new-age solutions roll out and mature, the ripple effects will undeniably be felt across the industry. In a world where water management is often overshadowed by flashier tech narratives, Veolia’s strides remind us that the future of water is not just about conservation but smart, innovative, and incredibly efficient management.
So, as Veolia navigates the exciting waters of AI integration, one thing is clear: the company is not resting on its laurels but rather, surging ahead to shape the future of water technology. And that’s something worth keeping your eye on.
My Full Conversation with Andrew Collier (Veolia)
These are computer-generated, so expect some typos 🙂
Download my Latest Book - for Free!
Antoine Walter: Hi, Andrew. Welcome to the show.
Andrew Collier: Thank you. Good to be here.
Antoine Walter: You were moderating roundtables today around the highly trendy topic of artificial intelligence and large language modeling. So I’m expecting you to come from OpenAI, but you’re from Veolia. So what’s the link?
Andrew Collier: Actually, we’re using some secure versions of open AI in Veolia, and we’re tackling a couple of different use cases that I was sharing today in those round tables.
First one is how do you make all of this domain knowledge that we have easily accessible and actionable within our company to improve internal productivity? Man, I don’t know if you’ve faced this, but sometimes there are Thousands of fact sheets and, you know, reports and everything kind of squirreled away in different parts of the organization.
How can you use a conversational bot to unearth that and make it easily accessible? That’s one area. Another area is applying LLMs to IOT data or time series data. Can you get a kind of expert assistant to look at that data, categorize the data, start to interpret it and feed some recommendations so you spend less time really in the diagnosis and more time in the.
Hey, I got to go solve and fix this problem for the customer.
Antoine Walter: So the first part I get it because I did it myself. I created what I called, uh, don’t waste water bot, which goes into all the transcripts of all the conversation I’ve had on the microphone. I vectorized it and then use GPT 3. 5 because it’s cheaper than four.
And then you can access that. So that I understand. And I guess at the size of a company like Verlia, you probably have a wealth of data sheets. So I see the use case. I’ll come back to the use case because it’s interesting. But I wasn’t expecting IoT to be a thing because LLMs are by definition language models.
So how do they perform? Behave when it comes to data crunching and maths,
Andrew Collier: you know, everything boils down to zeros and ones and can become semantics. And so there are ways to interpret data using natural language. And there are startups out there that are doing this very thing. Time GPT is one that you might come across.
How do you quickly, without writing any lines of code, figure out anomaly detection or forecasting? I think. A lot of this is going to become more pervasive and, and common in our area. We haven’t exactly figured out how to make that work within our context, but that’s something that we’re working through and we’re excited because it’s all kind of unfolding in real time.
We’re experimenting and we’re, we’re figuring out how do we apply our expertise where it’s really needed and get. A huge productivity lift out of these tools that can tackle a lot of the mundane, you know, day to day stuff.
Antoine Walter: So it’s internal stuff, or do you also intend in the long run to have it open to the external world?
Andrew Collier: Right now it’s all internal, but I hope soon, uh, if I can convince our legal teams to, to make it external. And one area where we think this can show up, uh, first is we have IoT platforms where we’ve got 50, 000 connected assets out there in the field. These are cooling towers, boilers, UF. systems. We know that, hey, an internal user or an external user is logging in.
They’re looking at this particular site. They’re looking at a ultrafiltration asset. They might be trending these parameters. Can we use that context to tee up the right kind of pre engineered, prompts for the LLM. So that if I’m asking you questions about membranes, it knows I’m talking about UF membranes, not RO membranes, because I’m looking at a UF membrane, or if it needs some kind of regional benchmarks, it knows where I am in the world because I’m already in that context of the user flow.
So that’s part of what we’re exploring is not just using chat GPT, kind of the general open to the entire web, but how do we curate a dataset and then make that available to the user in the context where they are in the user experience?
Antoine Walter: I’d go back to the depth of that. Um, sounds super interesting, but I think.
We need a bit of context. What is your role at Veolia?
Andrew Collier: Yeah, and Veolia is a big place, so maybe I can start with a big picture of Veolia for listeners, and then I can say where I fit in that big picture. Veolia, water, waste, energy, 218, 000 employees. I’m one of those. Last year, about 45 billion in euros.
About 41 percent of that or 18 something billion is water related. And in that water domain, there’s a water tech zone. And that’s about almost 5 billion in revenue as well. That water tech zone is a bit upstream of some of the utility services, but it’s creating proprietary products. These are the Z Weed membranes that you might know about.
Antoine Walter: Let’s be specific. That’s Veolia Water Technologies and Solutions?
Andrew Collier: Ah, no. So it’s, it’s partly, uh, Veolia Water Technologies and Solutions plus Veolia Water and Technologies, which is the legacy Suez business, which also came from GE Water. That’s where I, I came through the organization and the Veolia Technologies business combined into this WaterTech zone.
So we together operate pretty nicely. We have good geographic counterparts in different places and we do a couple of different things. We design and build. Plants, right? We manufacture proprietary products, whether they’re again, membranes, chemicals, separation technologies, and we provide services. And we do that mainly to industrial customers.
It could be in food and Bev. It could be in HPI, CPI, microelectronics, pharmaceuticals, but that’s the business we’re in developing these products and then providing them mainly to industrial users.
Antoine Walter: So the five billion chunk within the 18 billion chunk
Andrew Collier: within the 45 billion. Okay, so that’s
Antoine Walter: where we are right now.
That’s
Andrew Collier: where we are. And I’m part of a digital team. Okay. In that water tech zone, and we are responsible for bringing I. O. T. Solutions to our customers again. This is helping them get data from the field to the cloud to a visualization platform to do basic alarming, trending, reporting, those kind of things.
But then we take that data and we have. An analytics applications that we put add on. So for some of those core assets, whether it’s a cooling tower or it’s a reverse osmosis membrane where we want to forecast the time to clean the membrane, or maybe it’s a condenser in the power industry, and we’re trying to figure out how fouling is going to impact the power generation of that unit.
We’re using analytics to figure out how to optimize those systems and where we’re going next. And what I’m really excited to talk to you about is we have a new product that’s coming out. pretty soon. It’s actually out, but we’re announcing it in the next couple of weeks. It’s called HubGrid Water Footprint, and it’s more of a plant level view of sustainability impact.
So historically, we’ve been pretty asset focused. That’s our core understanding and our core domain. Uh, but this is up leveling to a different persona. Maybe it’s that ESG leader. Maybe it’s that sustainability director who’s looking either at a plant or a, Uh, a fleet of plants and is trying to understand, okay, what is my water footprint?
How do I build a water flow? So I can understand the water coming in, where it goes, where it goes out. So kind of a mass balance of water. How can I do that for energy? How can I look at greenhouse gases? How do I understand my risk? So, you know, I I’m operating these plants in these parts of the world. How can I layer in the water basin risk or the plant risk?
And then how do I start to build a list of projects, you know, that are net positive projects. So we’re very good at optimizing again, cooling towers. So can we work with the customer to lower their overall plant? water footprint by figuring out how to optimize a cooling tower. They might work with another company to optimize another asset, but they’ve got the whole view now and we can help them drive down their sustainability, you know, impact.
Antoine Walter: Let me see if I get that. You would have then upgrade water footprints, which is scattered to that ESG executive and he can have a portion of that then still within the Vilda tool, which does. again, build your stuff, but also third parties, which could then plug in. And so you would be kind of building the platform on which people would then integrate.
Andrew Collier: Yeah. The first step is really giving transparency. Like, where are we today relative to our water footprint and our reduction goals? And then we get into, okay, how do we improve that and whether we can help improve that or they take that information and can improve it themselves because they have now sensors that are automating the rollup of those water figures.
At least you start with the transparency.
Antoine Walter: Doesn’t everybody want to be the platforms these days? So you’re on a very, very competitive marketplace.
Andrew Collier: Sure. And. And, you know, a lot of these things can plug and play with each other. Part of the challenge is getting the data to come together in a way that you can make decisions.
And actually back to some of those LLMs, whether it’s the visualization platform that we provide, or whether it’s interacting with a chat bot that is tied into those various data systems, if we can understand, you know, the current level of performance. And then start to get some recommendations on how to improve that.
It doesn’t necessarily matter which pane of glass you’re looking through, but we’re trying to figure out how do you bring all that data together to make better decisions.
Antoine Walter: Upgrade Water Footprint is about to get released. And as I’m super slow with editing, chances are it is released by the time you’re watching this, but how are the different bricks insides?
Ready? What do you build? Do you release something which is like one on person all featured all in? Or is it like not a beta, but a first step, which then will evolve?
Andrew Collier: In software development, we’re constantly evolving the product. We work in an agile fashion. So every two weeks we do a sprint and we release new features.
This you would say is our kind of first beta version. So it’s got all the features that you need to do what we just talked about, but it will evolve over time. And the way that we’re kind of road testing it is we’re actually deploying it in 15 to 20 of our own plants this year. So again, we manufacture membranes in India and North America and Europe, pretty much all over.
Actually, our EHS team came to us and they said, Hey, we’re wrangling all this data from this source and that source and trying to roll it up in Power BI and Tableau. And what we really need is this. And we said, Oh, we’re working on this product here for customers, which is going to do that. And they said, Oh, well, let’s start using that internally.
And that’s going to help us really make sure that it can scale and then take some of those proof points to the customer. So we’ve eaten our own dog food, as you will, before we. We can, uh, we can scale it up. We do have a, you know, a customer in North America that’s, that’s using this now and the way that it typically works is we’ll go to the site, we’ll walk the site and do a plant audit and we’ll make this water map, you know, all the flows.
It’s that mass balance of the water footprint. We’ll tag all their various assets. So we have a clear sense of how water flows through the plant. We’ll add hardware and sensors if they don’t have flow meters here or there. So we can kind of sub meter throughout the plant. Then we go through kind of a diagnostic again, looking at the risk profiles for that particular.
And then it starts kind of working through the ideation of where are these net positive projects that we can work on together.
Antoine Walter: You just dropped some words which sound like modules of Hubgrade. So I’m just going to check. Plant Audit is what you’re marketing today as Hubgrade Plant Audit.
Andrew Collier: Let me just say that Hubgrade is a Veolia brand for digital products.
So Hubgrade really is a portfolio that covers water, waste, and energy. And there are several different products in that portfolio. We do have one called Hubgrade. performance plant, and that is specific to mainly municipal wastewater treatment, although they can do industrial water wastewater treatment as well.
And that’s actually a pretty impressive product. It’s been around for quite some time, and it’s, it’s one of our more advanced products that does a complete automation of a wastewater treatment plant with AI models to really dial in the biological health and improve the performance of those plants. If in that flow sheet and where, you know, there is a wastewater treatment plant.
Plant in that broader flow sheet, we can deploy hub grade performance plant as one of those opportunities to drive net positive impact. Let’s say there’s a condenser tied to a power generation unit. We can deploy one of our analytics called empower that helps on condenser performance. So it’s more about getting that plant level or fleet level view, and then picking all these various different tools that we have in the toolbox to try and make an impact in different places.
Antoine Walter: An impact for you is net positive.
Andrew Collier: Net positive. Yeah. And that’s really how do I lower my water usage? How do I lower my energy usage, my greenhouse gas emissions, helping the customer do that. In fact, that’s kind of how we run our business. We have a process called the value generation process, and it’s really simple.
We need to prove to you why you should keep us around and whether it’s how our chemistry unlocks economic savings or environmental savings, or whether it’s how this equipment implementation or retrofit can, can do that, we document that and we have our customers sign off on those saying that’s a real savings.
And that’s how we prove that, Hey. You should keep us around and we call it proof, not promises. And actually, if I can peek at my notes, I brought a couple of facts. These are relatively conservative because this is just in the last 12 months, what we’ve had customers sign off on. So there’s a whole pipeline of projects that we’re pursuing throughout the year.
And this is what comes out at the other end. So there are about 800 projects delivered and signed off by customers that represented 167 million of operational savings, some of which came through these environmental. savings. 240 million kilowatt hours of energy saved 14 billion gallons of water saved 46, 000 metric tons of waste reduced and roughly 18 million metric tons of CO2 equivalent reduced.
So it’s just
Antoine Walter: that’s for 2024.
Andrew Collier: That’s for the last 12 months. Okay, rolling 12 months, rolling 12 months. And we keep track of that. And that’s how we try to, uh, measure our impact. So net positive is what can we do for our customers in aggregate? And then how can we help them achieve their sustainability goals?
Antoine Walter: You want your customer to keep you around, but you’re water tech. So keeping Veolia around means Veolia operations. So that means that water tech has one of the missions to feed into helping Veolia operations to. Keep their customers happy. Do I get that one right?
Andrew Collier: Yeah. Again, we, we tend to make things a little complicated in the world, but, um, we have different business models.
I mean, we can be providing technology to the regional municipal businesses to like a drinking water plant or wastewater plant can be operated by Veolia North America, but the technology can come from the water tech zone. Or if it’s an industrial customer, we can typically give those services directly.
So civilian water tech would in industrial cases also be an operator. Yes. Yes. Interesting. Particularly in, uh, on the chemical side of our business where we’re monitoring equipment, again, cooling towers, boilers, whether it’s in process industries or you’re using water as an ingredient in food and Bev, or you need ultra high, pure water.
We are designing those flow sheets. We’re typically using the products that we manufacture in those flow sheets, and then we can implement and then operate those plants as well.
Antoine Walter: The product development is the core of. And all the different branches and all is amalgamated and came together. I fully get that.
What I’m a bit more surprised is the layers of development of AI LLMs platforms is not necessarily the core knowledge of Veolia. So does that mean you develop it internally and you are building up that capacity because you say that’s going to be core in the future? Or do you outsource it? that work and eventually do external growth or different name to M& A.
Andrew Collier: I’ve been impressed with the speed with which the Veolia group has been diving into LLMs, right? They, they created a secure version that we can use internally. They published an AI policy very early saying, here’s what you should do. Here’s what you shouldn’t do. There are resources at the group level that we collaborate with.
We say, Hey, we’ve got this idea to tailor this, you know, database and make it available for UF users. How do we do that? And we use their building blocks to make sure that we’re in alignment with that policy and those, that architecture, but we can experiment and that’s, that’s really good. We do that with our internal team.
So we’ve, I’ve got a team, I have a digital center of excellence that’s made up of product owners around the world. We have a bunch of different. Software development teams in Bangalore, India, we have AI and data scientists in, in that team. So they’re doing amazing things with, with these tools and they’re experimenting and trying to realize these use cases.
In some cases, we also partner if we don’t have particular capability or know how, but we can layer it into our product and offer something that is better with a combination of different things. We’re not tied to just trying to figure out everything internally. What we’re trying to do is take that internal domain knowledge.
That is our secret sauce and then pair that with these tools to make things more accessible for our users.
Antoine Walter: You mentioned your data analysis slash AI team. Did it exist five years ago? It did, but it was like two people
Andrew Collier: and it was as much smaller and it was kind of process engineers turned amateur data scientists and we’ve grown it up over the last several years
Antoine Walter: to how many people?
Andrew Collier: We’re about 50 people, which is, it’s not just data scientists. I mean, that’s everything from what we call our team. tech dev team, which does data science. And they, they know machine learning, computer vision, natural language processing, gen AI, it includes our app dev. So that’s front end development, backend development, and it includes kind of the product owners that own these different things.
And that’s not just for our IOT platform insight or these analytics or hub grade water footprint. I’m also responsible for some software applications that we use internally. So we give our commercial and engineering teams tools to do value calculations. So Like I was just mentioning this value generation process, we have a suite of tools that help you calculate what’s the greenhouse gas, you know, emission reduction potential of this particular thing.
Or how do I pick the right chemistry for this particular water challenge? We also have design tools. So, okay, I’ve got this water challenge and I’m building a UFO. flow sheet. How do I use software to make that design process faster, more efficient, more standardized so that we can move quicker. So that team is supporting many things, uh, and they do great work.
Antoine Walter: It’s not a big secret that I’m a former Suez guy. I’ve said it countless times on that microphone as I was in a very, very large team of one being my own boss and doing my own calculations and offers and stuff like that. It’s not that long ago and still, it feels like when I listened to you, it was I mean, my tools were a series of Excel sheets, not even one single Excel sheets.
I had different Excels to do different stuff. And then it was a lot of handworking. And now you’re telling me that internally you’ve built something which resembles what Transcend were describing on this microphone. So some of the coolest startups out there in the field. You’re doing similar stuff internally to Veolia.
What’s the magic trick?
Andrew Collier: We’re a big organization. We still have a long way to go, but we’re trying and we’re, we are moving the needle and things have accelerated, I think in the last, you know, a couple of years. Look, I,
Antoine Walter: but what’s, what’s the trigger? Because exactly you’re, you’re the leader by 10 miles.
Like you’re, you’re three times bigger than the second largest water company. You have all the reasons. To be conservatist to not touch anything and just to keep doing what’s and it’s the opposite that’s happening. So I’m trying to understand why
Andrew Collier: I think people are realizing that this whole generation revolution is a complete step change and it’s not something that is a progression of incremental improvements over time and the swiftness of the response and the attention that I’m giving credit, you know, to the to the corporate team Villa has taken to give this the right attention.
I think. Is changing a bit of the way that we tackle these problems. I still have challenge of prying Excel sheets away from engineers, you know, but we have taken a real hard look at how do we keep a very user centric approach? Like if I build the best software application and it stinks, they’re going to keep using the Excel sheet.
So we really need to figure out how to keep the users in the process of development. So again, every two weeks we do a sprint demo and we bring the users and say, is this meeting your needs? No. Okay. Then we’ll, you know, keep, keep working at every two weeks. We still have a long way to go. I mean, uh, these are, these are not things that happen overnight, but compared to a couple of years ago, I think we’re in a much better place.
And you asked me a few years down the road. Hopefully we continue.
Antoine Walter: I’m curious to understand how you do that. You’re from the U S so you’re going to have. The reference, I forgot the name of the city. It’s a city which is next to the great lakes and which is supposed to be the one where you test your stuff before rolling out because it’s the one representative of the demographics of the U S and if it works there, it’s going to work.
I’m going to find it. So I have it somewhere in, in, in the mess that my brain has become after two days of Bluetech forum. He turned me as in for the safe. It was Columbus, Ohio. I was referring to where I’m heading with that is when you say every two weeks, you’re presenting it to your users in a 220, 000.
People group, it’s not like a town hall where you get everybody on the call and you say, Hey, let’s test it. So how do you select your beta users, your early adopters,
Andrew Collier: a couple of different things. So each of our tools has a product owner and that product owner is responsible for engaging with the various stakeholders, whether they be internal users or external users.
So there are conduit to our users and they will again, run. Beta user groups getting feedback before we launch something. They will survey those users on a regular basis to try and get feedback, or we’ll put up a actual ways to engage with the user in the application. So, you know, they’re going through and there’s a pop up and it says, Hey, you want to learn about our new membrane, uh, analytic, you know, click here, or there’s a new feature and we can kind of understand how they’re engaging with things and we give them clear channels if they want to, to engage with us and give feedback and we’re trying to be user, user, user centric, and then we’ve been really Honing our, our agile development approach over the last few years, whether it’s the way that we manage a backlog of things that we need to do, whether it’s the way that we break that down into bite sized pieces for our development team to pick up and run with, whether it’s the way that we engage at least the product owner and some stakeholders as a proxy for the users.
On an ongoing basis, those things again, didn’t happen overnight. So we’re, we’re trying as hard as we can to model, you know, the best software companies out there, but do it in our own context, you know, do it by pairing software developers with process engineers or chemical engineers to figure out, you know, how do we make the day in the life of one of our users, internal, external, better tomorrow than it was today.
Antoine Walter: I’ve spent far buzzwords like machine learning AI. And the one thing which comes back usually when those episodes pop out is, Oh. God, buzzwords, why are they using that? Do you have metrics? Like, I’m a Veolia Watertech zone or sphere. Do you have a better name for that? Watertech
Andrew Collier: zone. Watertech zone.
Antoine Walter: I’m a Veolia Watertech zone process engineer.
I’m preparing a design for a proposal for a wastewater treatment plant. I actually don’t want to get rid of my Excel because I love my Excel. But you want to convince me that you have a better tool to help me. What are the benefits for me as a process engineer to hop on what’s surely the future, but not my present?
Andrew Collier: Yeah. I mean, there’s always an adoption curve, right? You get some early adopters who are lean into technology. They see the potential, they want to be part of the process. And it’s always nice in those first engagements, cause you think you got something. And then you hit the fast followers. You know, you got to convince after some of those early adopters have shown them, Oh, I did this.
at a fraction of the time or, Hey, look at this output. I only had to do X, Y, and Z. So this part of it was automated. So you can use the early adopters to convince the fast followers, but it takes a long time to get that late majority to, to kind of move in your direction. But
Antoine Walter: what’s
Andrew Collier: the
Antoine Walter: aha moment? What is the, the specific tool in the suit where they say that is so good, I will need to convert to that.
Andrew Collier: There’s no secret, you know, it’s, it’s a matter of constantly working with them. To figure out so I don’t know if you’re familiar with the value proposition canvas or some there’s some kind of elements of design thinking where forget about features and functions for a minute, just focus on this is my day to day job and these are my pains and my gains, right?
Like this thing is annoying and this thing makes me happy at the end of the day. If I can help on either of those dimensions by building this feature. Or, you know, rolling out this new tool, then I will get, you know, adoption and I focus on user adoption on all my tools. And it’s not just Hey, did they log in once during a training session and never come back?
Like we look at how many return users do we have month over month? And for the tools that don’t have many users, I stopped supporting those tools. And I go to where the need really Really is. So it’s, I don’t know if that’s answering your question, but it’s a,
Antoine Walter: it does. Uh, when you say you look into it, does that mean that’s your KPI?
Like you are measured or measuring yourself on the adoption of those tools and try to kind of customer success, but internally.
Andrew Collier: Yeah, absolutely. Yeah. We are trying to be user obsessed and, uh, and measuring user adoption and traction is a key part of that. And when you talk about metrics and how we hold ourselves accountable, we also within the digital team, we use.
A framework called the Objectives and Key Results Framework. OKRs. I don’t know if you’re familiar with that. And again, it took us a while to kind of get our arms around that, but we, we think about what are the quantifiable impact that we want to drive. And for us, that’s driving revenue through these digital solutions that we monetize.
It’s driving stickiness through the engagement and those, you know, user adoption tools. We have milestones that we want to hit, and then there’s, you know, different things that we use to enable that. Building up some of our commercial function or releasing new headline worthy. Product features around Gen AI or changing the organizational mix by bringing in these kind of talents, you know, that helps us enable those kind of outcomes.
So we as a team rally around that, and we are typically looking at how many users do I have? How many sites are under subscription? How many assets are connected? You know, how much data is flowing each day? If I make the users happy, those other things are going to go up
Antoine Walter: when you refer to users. Those are your internal users, or do you
Andrew Collier: both?
We look at both internal and external.
Antoine Walter: Okay, so you you look at both. If you have a split, rough splits between what you’re doing, which is customer facing and what you do, which is. internal facing. What’s the ratio?
Andrew Collier: Yeah, it kind of is product dependent. I mean, for our flagship product insight that’s been around for a long time, it’s like 60 40 internal external.
But for hub grade water footprint, it’s going to be much more external focus because we are building that for the customer sustainability director who needs to hit these particular targets. And it’s giving that transparency. And then we bring the rest of the things, you know, together. So it depends on on the product.
Antoine Walter: There are so That could be a five year in the past question, this customer’s sustainability director didn’t exist five years ago. It’s crazy the pace at which you’re adapting to something which didn’t even exist in the market. So it means you’ve reacted to something which was transitioning in the market, you developed it, and you’re ready to go to market yourself as a company, which arguably would have all the right arguments in the world to be slow.
I’m still trying to To, to make myself believe all of that.
Andrew Collier: We’re trying to, um, anticipate the needs. And I think that I spent four and a half years in France and I got a flavor of sustainability and the seriousness that Europe takes on this topic. And we see things like the corporate sustainability reporting directive, that’s going to drive a bunch of companies to disclose, and they need to have data and good means of rolling these things up that are going to be a big driver.
We see the way that AI can. Help in terms of productivity, and it’s not necessarily coming up with something out of left field. Here’s my, you know, a I digital twin, but it’s actually embedded into what we’re doing. Like we already monitor 10, 000 cooling towers. How do we augment how our team can drive performance of those cooling towers by using things like forecasting the performance, identifying anomalies and then Giving recommendations to the operator on how to improve that.
Antoine Walter: Would it be too much of a shortcut? If I say that the number one asset of your upgrade, what a footprint is disclosure in one click.
Andrew Collier: That’s a nice line. Maybe, maybe we can license that from you, but yeah, I think it’s, uh, I mean, we, we think about transparency, resilience, and net positive. Those are kind of the modules of.
of that. And whether it’s one click or a few to get started, yeah, I think it helps you with those disclosures. It helps you with that transparency. And we’re all going to need to do more of that going forward.
Antoine Walter: Now I have to be the party pooper and the devil’s advocate and the old guy. You’re using LLMs to give you technical advice when they are known to hallucinate.
Come on, they don’t tell the truth and they’re so wrong all the time. You’re connecting so many assets. But cybersecurity, I mean, we have the Olympic Games on the horizon and we see that, uh, Russian bots are spreading fake news about France. So what could it be about cooling towers? I mean, I’m of course, yeah, yeah.
Making it far over the top, but how do you tackle those somewhat legit fears that people may have, especially if they’re from a different generation?
Andrew Collier: Yeah, yeah. They’re legit fears and we can, we can definitely talk through them. You know, I think in terms of hallucination of a LLM, one of the things that we’re focused on is we, we don’t just pick up.
Chat GPT and take the advice that we get from the general worldwide web. We curate knowledge bases that we already have vetted and we know, you know, is good sound domain knowledge for our organization. And we cater those to the chatbots. And then we go through validation process. That’s what we’re doing right now is we’re building these up.
We grab our process analysts and we say, Hey, go ask this thing questions and let us know if, if we’re getting good responses. And then if we’re not getting good responses, how do we continue to fine tune it? And we have found things that we need to fine tune. So it is about,
Antoine Walter: I’m going to be nerdy on. that one.
Sure. Fine tuning, training. What’s the type of thing you’re doing? Are you doing your own model? Because if you only could the size you have, you could do that. Or are you taking an off the shelf GPT, whatever, and fine tuning or even vectorize some data and add it to its knowledge?
Andrew Collier: It’s more on the vectorization of data and curation of the data.
So we can, one, curate that data set to what we need to be in there. Like, for example, if there is any sensitive information in that data set, We can scrub it, you know, before putting into the data set that way you’re not getting a Q and a and it’s serving up sensitive information. And that’s something that actually has nothing to do with the tech.
It’s not like going in and retraining the actual large language model. It’s about curating the content, which you apply the model to.
Antoine Walter: So actually, the biggest hurdle and the Big chunk of work you have to do is clean your own data and turn it into questions and answers so that you can feed your model.
Andrew Collier: That’s right. And we take the help of our experts. So we take a process analyst and we ask it to, you know, engage with this data set and then give us feedback. And there’s ways that we can fine tune that model or fine tune that content to get the right kind of responses. And again, this is all things that we’re learning kind of in real time as these things unfold.
You also asked about Kind of general security. I mean, everything that we’re doing again right now is internal. It’s through a enterprise version of a GPT. So you know, all of our data stays within our domain. We’re not reaching out to the World Wide Web. And of course, in terms of data flows in any provider who’s who’s giving IoT services, we have a host of different providers.
Uh, IOT security measures in place, make sure the data flows are, you know, secure.
Antoine Walter: By the episode two of this podcast, I had Veolia as a guest to discuss digital solutions and upgrades. And one of the things we discussed at the time is, would you at some point go to a fully automated plant? And the answer, At the time was no, you would go to a highly assisted and augmented human that has all the digital tools to support his decision makings.
But in the end, the one taking the final decision is still the human. But that was all long before GPT 3 came out and kind of put AI in the middle. mainstream discussions. I’m not trying to ask you if we’re reaching the AGI and, and, and hence, uh, I can do stuff alone. But we all know that technically speaking, water treatment plants, most of the water and wastewater treatment plants could be running alone for 10 years.
We just don’t dare to do it because of course there’s a risk element and there’s something to manage. But with this new advancement, which you’re bringing in, Merging all the possible buzzwords, machine learning, IOT, artificial intelligence, training, vectorization, whatever you want to call it. What’s the answer to the question change of will we soon or one day have automated plants, which run human less?
Andrew Collier: I think it’s a good thought experiment. You know, we should start from a position of what would it take to have a humanless wastewater treatment plant, potentially in, you know, remote locations or things like that kind of work backwards to solve some of the problems that we face today through automation.
Do I think that’s coming around the corner tomorrow? No, I think that a lot of these tools are still helping to augment the human operator. They’re pairing expertise with efficiencies. And I think we can fine tune things. Tremendously using these tools, but I think we still need some of the human, uh, element in there.
You talked about hub grade performance plant, and we’ve found that, you know, even in a really well operated plant, if you pair an operator with some of these models, you can still reduce energy by something like 38%. We have a case study that, you know, that’s
Antoine Walter: significant. I know the case study and I’m just smiling because you say well operated plants, because yeah, of course, but how well is it really operated when you still have like 40 percent of.
Sure,
Andrew Collier: sure. Yeah. I don’t know if this is a good analogy, but I think about it like, say you’re driving a car and even if you’re on a very straight road, you know, you could only touch the gas or the brake or the, or the wheel every two minutes, you’d still probably run into some issues or, you know, eventually, right?
That’s a good one. So it’s, it’s not a question of, are they getting the job done? It’s how fine tuned and how precise can you be in all near real time to dial it in to a higher level of optimization.
Antoine Walter: About six months ago, I had Kendra Morris on that microphone and we discussed about what Veolia was rolling out, especially in North America, to train a new generation of operators because of the silver wave and how that was kind of Veolia’s move to the market because they were not like binded to Veolia.
They would be then able to work for whoever, but it was a way to train the workforce to have people taking those roles in the industry, which is known to. Have difficulties to staff itself. If I now put that together with the vision of AI and digital tool, I’m going to enhance the human and make the work somewhat easier, somewhat less of a burden.
I have two ways to look at that. The first is to say maybe I need less humans in the plant, and it’s not like it’s taking the jobs of people because people don’t take those jobs. So it’s just solving an issue. The second is to think of what, uh, Meyer, Eric Simmons said when he was on that
Errick Simmons: microphone, that a lot of folks get in the tech industry, but if you’re going to digitize and rethink water, just like folks have iPhones and Samsons, we need people to be thinking about water and wastewater at the same scale or even greater,
Antoine Walter: we need to make the industry as sexy as working for Samsung or Apple.
And if all of a sudden we are also playing with LLMs and AI, we are as sexy than Samsung and Apple, because on top of that, we have the purpose that it’s. Water centric. So which of the two would be the number one perk you would see?
Andrew Collier: You made me think of many things there. So let’s start with the fact that I see this as helping some of our workforce challenges in a couple of different ways.
One is we have a lot of folks retiring and we’re trying to bring in new talents and get them up to speed quickly. If we can capture some of that domain expertise and offer it up to some of those new employees through tools like LLMs, can you get them on board faster? Yes, I think so. Can we now Start harvesting the knowledge of the people who are going to retire in the next couple of years and build that into our knowledge base.
We have to start doing that now. I look in our company and we have Google spaces, you know, where we’re doing Q and A’s with all these experts. Hey, do we have a reference for this? I’m trying to solve this problem. What do I do when this happens? And it kind of, it’s transient. Like it comes into this space and then it, you know, it goes away.
How do we take that Q and A, which is already validated? by our experts and then package that into the knowledge base of these LLMs. If I restrict the LLM to a certain set of documents, it won’t know the answer if it’s outside of those documents. So if it can’t answer the question, but then we can go query an expert and then add that to the knowledge base, that’s going to just enrich that over time.
So I think we need to do both of those things on the retiring side and on the new entrant side in terms of the workforce.
Antoine Walter: So you could have in the future a new hire, which is a human, but trained by AI to start with. AI trains the human so that the human can do his job, which is kind of a reverse way to look at it.
Andrew Collier: Yeah, that’s a good way to look at it. The other thing I’ll say is somebody meant said this during one of our roundtables earlier. They said, you know, AI is not going to steal your job, but someone who knows how to use AI will steal your job. So all of us need to figure out how do I become a more productive worker using these tools?
Otherwise, somebody is going to become a more productive worker and come along. So it’s not like, hey, the machine is taking my job, but the machine can augment it. The human and that’s where we need to you know i’ll get to
Antoine Walter: your preaching to the core here i mean people sometimes ask me how do you do to run the podcast as a side to my daily job which is a full time job well i’m using i counted twenty six different eyes so it still needs a lot of work.
But I could not run the podcast in its current shape with the tools that existed 10 years ago, it simply would not be possible. So I feel the element of augmented, but there are not so much stakes if my podcast is not produced well, why it’s not produced well, doesn’t change the word. Whereas if your water or wastewater treatment plant doesn’t run well because of a glitch.
It’s might go to trouble. I’m just trying to not really not want to be the party pooper because I’m super hyped by everything you said so far. I’m just trying to, to anticipate the legitimate fears which come with all we’ve discussed so far.
Andrew Collier: And that’s why I think humans aren’t coming out of the mix anytime soon.
We’re probably not going to have pilot lists. Planes tomorrow either, but a plane can fly itself, you know, around the world. So I think you still need a bit of that control and these things are evolving, you know, so we need to take it a day at a time. Yeah, I, I try to keep things pretty simple. I build something, I ask for feedback.
I iterate on that. And I think these kind of, you know, things that are reshaping the world around us, we will get lost if we try to imagine all the different options that can come out of that. But if we take it, you know, keep experimenting, keep getting, being hands on, I think we can learn how to really make, make this useful for us.
Antoine Walter: We will get lost, but I’m still going to ask. I want to do futurology now. The next thing which comes out is upgrade water footprint. That’s again, because of my slowness, probably already out by the time. People listen to that, but what’s next in the pipe?
Andrew Collier: I’m releasing some kind of augmented visualization dashboards for our core assets, cooling towers, boilers, UFRO.
And I’m, I’m doing this in a standard way that will allow me to start do more comparative analytics across all these sites. So we anonymize, you know, their performance, but because I’ve got 10, 000 cooling towers, I can tell you if you’re in this industry with this kind of cooling tower, are you in the upper quartile or lower quartile?
Then I want to, you know, continue to augment those with these kind of analytics that do forecasting, anomaly detection, recommenders. Those tools again, should be complimentary to upgrade water footprint, because as you’re creating sustainability goals and you want to drive those net positive impacts, well, here’s some of the tools that we can use to do that.
So I think in the short term, it’s, it’s kind of enhancing some of the capabilities that we have today. In the medium term, I think, uh, taking. This idea of, of LLMs and creating these kind of virtual process experts. So it’s, it’s kind of co looking at the data with me. That’s what we want to make external facing.
We’re still got a little ways to go there. I think longer term, we want to figure out how to use generative AI for generative design. You know, we have solved so many different water treatment problems. So we have databases of water qualities and then the solutions that we built, the flow sheets that we designed to solve those problems.
How do we You know, use all that information to make an engineer’s job so much faster and so much more dynamic so that they can go through so many iterations of different plant design, but then start optimizing, Hey, I want to optimize this for cost, or I want to optimize this for carbon, or I want to optimize this for, you know, this parameter, the other parameter, how do we use these tools to get to much better designs and then keep that model all the way through to operation, these things start to circle back, you know, so if you can Connect design to operation, to the, you know, data visualization, to the decision making and then use the learnings from that to reinforce the design.
You know, that’s the ultimate Holy grail.
Antoine Walter: Lots of things. First, it’s interesting because it sounds to me like you’re reversing the leader’s curse. The leader’s curse, as it said, would have meant that you would be a bit more waiting and on the defensive side, but you’re not, you’re taking it to the other way, which is if you have 10, 000 cooling towers, which are connected, you have much more data than anyone else.
Let’s use it to your advantage and let’s do it. Data crunching. So that’s, that’s already interesting in itself when it comes to this generative design and already namedrop transcends earlier in that conversation with the natural way for Veolia to keep developing stuff internally, or would you just do a move and acquire transcends?
Andrew Collier: And we’re actually working with transcend. They have, I thought you
Antoine Walter: were going to tell me about acquiring transcends. Ah,
Andrew Collier: good. A scoop. Yeah, they’ve incorporated our MABR design into their platform and we have a product called the Zilong and it has a bunch of benefits and kind of alternative. to MBR or conventional wastewater treatment technologies.
We’re experimenting with them. You know, again, we have the same goals of making the user experience and the design process more iterative, more user friendly, more productive. And so we’re, we’re working together to exchange ideas and we’ll see, you know, we’ll see where that evolves. We’ll
Antoine Walter: see. Is it not yes?
Is it not no? It’s we’ll see. Okay. I keep wanting to, to go back to me 10 years ago. and say don’t worry it’s awful now but it’s gonna be a lot better in not even 10 years ago like six seven years ago which sounds like an eternity now it makes sense to scratch your own itches because everything you discussed so far is stuff i can see myself having those struggles so i i get where your user feedback comes from that is what’s next so Having potentially generative design, having stuff integrated.
But if you look now on the horizon, like five years, 10 years, what are you building?
Andrew Collier: That’s a hard question to answer because, you know, five years, 10 years ago, would we have imagined, you know, where we are today? I don’t think I have a real crystal ball for five, 10 years for you. I do know that the way that we manage data today is probably going to feel so careless in five to 10 years because just.
The requirements for reporting, you look at like the banking industry or, or other sectors that are so data focused. And then you try to explain to them how we, you know, cobble together data for different things. I think we’re going to have an immense growing up and maturity curve, you know, to, to get really sharp with.
Environmental data from a reporting and disclosure perspective. And we’re going to put all these tools, that’s going to be the pain point. And all these tools are going to help address, you know, how do we do that in an effective way? No big, like 10 year from now product reveal, but, uh, you can welcome me back in 10 years and we can reflect on.
I’ll welcome you together
Antoine Walter: with your virtual assistant GPT 7 and, uh, I will have a lot of fun. So on a personal note, you shared your IKIGAI today. I learned that we share to be second generation water professionals. That is a human touch, which again, if the, if I’m the party pooper here, kind of disappears with, yeah, it’s AI, it’s, it’s job like another, which is exactly the opposite from what I said 10 minutes ago.
So see how I’m logical. What’s the human connection? What drives you personally in developing a hub grid, water footprints or generative design?
Andrew Collier: At the end of the day, you know, sometimes I take a step back and I look at the work that I’m doing and I, I am very curious still about the science behind it.
And then I think about if I were doing another job outside of this domain, that how bored I’d be so quickly, you know, because I really find the environmental components or, you know, the technology, how it works to separate, you know, these constituents. I find it really. Interesting. And so I geek out on that level and my training, I’m not an IT professional by training.
So this whole digital journey, which started several years ago, you know, has been a fun dimension in the water domain. So as long as I’m continually learning new things, whether it’s on the commercial side or on the digital side, but it’s still tied to the environmental domain and it has a bit of water thread into it, then I’m, I’m still motivated, you know, and, and that, again, I shared the story about how my mom, you know, Was executive director of the Delaware River Basin Commission, and she was responsible for managing a watershed for 15 million people.
And, you know, my dad, he’s a land use planner, but we grew up through engagement in the Boy Scouts and, you know, had a connection to sustainability and community. So those are still part of my core values. And if some of that shows up at work, then that’s the, that is the kind of icky guy, you know, that you’re going for.
Antoine Walter: I don’t want to take you down a sidetrack. There’s so many I wanted to, to, to open here. But, um, yeah, I connect to what you’re saying. So it’s a, it’s a shitty conclusion, but it’s fascinating. Andrew, thanks a lot for everything you shared in that deep dive. I have a couple of rapid fire questions to round it off.
If that’s fine with you.
Andrew Collier: Let’s do it.
Download my Latest Book - for Free!
Rapid fire questions:
Antoine Walter: What is your definition of innovation in action?
Andrew Collier: I think touching on Some concepts from before, for me, it’s rapid prototyping and constant user feedback.
Antoine Walter: Rapid prototyping, because, yeah, I didn’t open that door, but I noticed that you have, on top of all the other cool aspects, you describe, you’re applying design thinking, design thinking, agile, all of that, how does it come into the DNA of a company like Veolia?
Andrew Collier: It doesn’t show up automatic, you know, we have to build that, you know, through training, through practice, through looking at others, and, um, I’m a very visual person. So, you know, if I have an idea, I’m going to draw a picture and most of the time it’s going to be the wrong picture, but you’re going to tell me how you think about it.
And then I’m going to draw another picture and then I’m going to keep going, keep going. And that’s what we do every day in software development. Like we have an idea, we make a terrible mock up in, you know, Google slides. And then we hand it to a, you know, a UX developer who makes it beautiful. And then we show it to a professional and they say, that doesn’t make any sense.
And then we start again, you know, and we do it again. But we, we eventually get to something that is useful versus just staying kind of in our corner, thinking we know the answer, spending a lot of time, a lot of time, a lot of time developing saying, ah, here it is. And then they’re like, I don’t need that.
Antoine Walter: What is the most innovative water solution you’ve seen in the past 12 months?
Andrew Collier: You know, that’s a tough one because even in this, conference, I just met a bunch of really, um, inspiring startups. I do participate, um, as a judge and imagine H2O and we’re a sponsor of them. So I always get to see a nice pipeline and actually one stuck out that I was judging and it has nothing to do with.
You know, what I work on every day, they’re a company called sample serve. And what stood out to me was they’re part of the 2024 cohort of imagining show. If you look at some of their marketing materials, they’re so good at getting to the pain point of what they’re addressing, which is environmental sampling and making that process easy and scalable.
And they do it in a way that is almost kind of comedically engaging. You know, they kind of paint a picture of, is this really how you do it today? And then this is what it could be, you know, with our solution that they just nail, you know, the kind of value proposition, which sometimes we totally miss. You know, it’s like, let me tell you how cool my technology is and all the specs.
You know, like, okay, but what problem are you solving? So they started with the problem and then they went through. And so that, that’s something that stood out for me.
Antoine Walter: Which brings us back to the value creation Canva, because that means you have to start from the challenge and then offer the solution. And you’re so right that we too often do it the other way around.
Absolutely. In one word, what is the biggest challenge facing the water industry today? Speed. Okay. More than one word.
Andrew Collier: Yeah. I mean, we’re, we’re inherently slow moving for good reason. Cause we are working on critical infrastructure and human health, but we tend to observe. Okay, what’s the energy sector doing?
And then we should do the same, you know, or what’s, what’s happening over here. And you hear it in the conference, you know, today we’re slow moving. We’re slow moving. So. How do we take a different tact? Maybe it’s just about finding safe places to experiment or safe places to collaborate, safe places to share data.
That seems to be a big, a big barrier, but we need to try and experiment faster. And hopefully the water industry can take some of that, you know, agile approach to solving problems.
Antoine Walter: Who is a water innovator you admire and why?
Andrew Collier: I’m going to use my IKIGAI again. My answer to this is my mom. It’s not that she’s a like a She invented a new technology, but the innovation that I respect from her was in the period that she was running this organization, she had to keep four states, you know, all in sync.
And then in Pennsylvania, where the Marcella shale underlies the basin, they were rolling out hydraulic fracking all across the state, but she dug her heels in and said, you know, there shouldn’t, there should be no fracking in the basin because we can’t risk having any contamination enter into the Delaware river.
Cause Hey, by the way, the Delaware river provides Half the drinking water to New York City, all the drinking water to Philadelphia, where I’m from. And so, like, you know, she had to really stand up to the political stresses of that, get everybody, you know, in agreement and figure out the right innovations from a policy side to keep what she cared about safe.
Antoine Walter: I love the personal connection and also love the policy innovation aspect, which is a vastly overlooked. What’s one single piece of advice you would give to emerging water entrepreneurs?
Andrew Collier: I think keep it simple. Imagine H2O typically talks about focusing on your beachhead market. You know, what is your most attractive market and just focus on that.
If that doesn’t work, pivot, but don’t try to do too many things. When I was judging some of those, you know, startup presentations, it was, we’ve got our first customer in the U. S. and now we want to go to the U. K. It’s like, why? Go get your second customer in the U. S. It’s like, stop. being distracted by other regions or other markets.
You know, you really need to secure yourself in that beachhead and then, you know, you can always do more.
Antoine Walter: How much time does it take to, to, to be like a, an imaginative judge?
Andrew Collier: Not that much time. I’d encourage other people to get involved. It does take a few hours to go through all the documentation, but you always learn something, you know, about the startup.
If you can be helpful, giving them a different point of view, then you feel like you’re helping them in their path. And then you get to go to fun Imagination2O events or fun, you know, blue tech events and meet them and, and, and it’s, it’s rewarding. So
Antoine Walter: what is one water taboo that you broke or you believe we should break
Andrew Collier: every single day?
I hear, Oh, we tried that 10 years ago and it didn’t work. And we just talked about how different 10 years ago was from today. And we just need to stop doing that. We need to challenge the status quo and try from a fresh angle.
Antoine Walter: What’s a common misconception about water innovation that you’d like to debunk?
Andrew Collier: Yeah, some startups fear working with corporates and I understand why. We can be It’s a bit cumbersome to deal with, bureaucracy, too many people, too many, you know, distractions. But I have been in part of successful corporate startup partnerships from the corporate side. And for me, you need to be upfront with what you want in the very beginning and try to set kind of a relationship contract at the beginning.
And if it’s not going in that direction, just be prepared to walk away and no hard feelings because I hate wasting people’s time. And these things can get drawn out. But if you go into it, you know, eyes wide open and at the very beginning, it can be very powerful when you combine the speed and focus of a startup with the scale of a corporate, you know, magic can happen, but I know it can be painful sometimes.
Antoine Walter: Look, I’m guilty of that. So I’m going to let you two minutes to make your case because often start to reach out to me and say, look, we have that product and we believe that it would be a great fit for Veolia. And we spoke with Veolia in. Insert the name of a country and uh, they think there’s a good fit.
I think that’s good If you take it, uh, we can then exit to verlier and everything is nice. I’m like don’t do that Don’t don’t first don’t go to verlier They’re so big and there are so many different entities You can get adopted in the country and that might still not move the needle in another country I might be wrong on that.
You might correct me on that one and the other one i’m mentioning to them is that, yeah, before they adopt you within everything they do, it’s going to take time. And if you’re starting from the roots, it will take time to go to the head. And if you’re starting from the head, it would take time to go to the roots.
So better have more than one path. to not even exit, but also scale. How wrong am I with that advice?
Andrew Collier: No, I think you’re onto many things there. You know, it is difficult. And at the end of the day, you got to figure out what’s in it for both sides. Right. So often a startup will say, Oh, I got a meeting with Veolia.
Like I’m going to 10 X in the next, you know, six months. And we’re so slow that we organize like two meetings in six months. Right. So you need to have realistic expectations, but then it’s a big organization. So we need to find the right people who benefit from the scale of this thing. So if you find the person who’s going to have an incentive to open this new market or scale up this new technology, or be more competitive with this, because that’s what they’re getting measured against, that’s, you know, core to their job, then you’re working in alignment.
Your incentives are aligned, right? Like you both want growth. If you’re just in the innovation department, whose incentive is, well, I got to screen 500, you know, startups and hand, you know, 10 down to the next business unit, they’re not invested in the same outcome that you want. It takes time, it takes attention and you might have to go through a dozen people before you find the one where your incentives are aligned.
But then when they are, it can be really, really powerful.
Antoine Walter: Xylem is a cool kid. I’ve had the discussion with them several times on that microphone. The fact that they have the Xylem innovation labs make them even more of a cool kid. Veolia is not a cool kid. You don’t have the Veolia innovation labs. Why?
Andrew Collier: We do have innovation labs and even, uh, Even in the broader Veolia for digital organizations, we have the birds kind of incubator and we have a couple of different things that get kind of treated outside of the big Veolia so it can have a little bit of oxygen to breathe. You know, I used to run the Suez Digital Hub and this was kind of an incubator.
So I’ve, I’ve lived that life and we learned many lessons along that way. Things like any financing of projects. Make sure that you co finance, you know, with a business unit stakeholder and a corporate stakeholder. Don’t just bankroll this with corporate money because it needs to live on after that. How do you, um, ensure that there’s a scale up plan within the business unit?
After the pilot, because again, otherwise you’re just kind of left with a proof of concept, but no owner, you know, to take that forward. There are pockets of open innovation, either in different Veolia business units or in the corporate function, but maybe it’s not as branded as, you know, Xylem Innovation Labs.
Antoine Walter: So if innovators listening to us would like to team up with you, where would you advise them to reach out within Veolia? What’s the right arm or the best arm to start with?
Andrew Collier: Yeah, there’s no one clearinghouse for that. So, um, yeah, again, it depends on the business line. Are you water energy waste? Are you kind of, uh, regional services?
So if I’m looking at a, you know, utility operation, I’m probably going to reach out through, you know, Veolia North America. If I’m targeting North America, if I’m looking at, Hey, I have a a product innovation, um, you know, you can reach out to, uh, we have an open, open innovation director for the WaterTech zone.
That’s a good start. You get, you get several entry doors. I was going to say, or we are connected to vehicles like Bluetech and Imagination2O and that’s a great way for kind of matchmaking to happen.
Antoine Walter: Makes a lot of sense. Yeah. What can and should I do for you?
Andrew Collier: You’re French, right? Is there a way you can help bring Picard to the United States?
That would be my number one. No, I think, uh, seriously, um, I think that we can sometimes get into a, like, catastrophizing mode. You know, here’s the doom and gloom of, of, uh, Of climate change, and you know, here’s the depressing stats about this or the other thing. I said in my Ikegai earlier, you know, I really, I think I’m a positive person, and if we can approach environmental challenges with a yes and or a what if type of mindset, then I think, you know, we can unlock a lot of potential.
That’s what I’d ask for your help propagating that.
Antoine Walter: Andrew, it’s been a pleasure to spend that roughly hour with you. If people want to connect with you to have more hours like that one with you, where should I redirect them the best?
Andrew Collier: Yeah. Best places LinkedIn. So feel free to connect, or you could check out our website, watertechnologies.
com to learn about our solutions.
Antoine Walter: So the website and the LinkedIn are in the show notes. Check them out. Thanks a lot for having been with me and safe trip back, I guess, to, to the other side of the pond at some point. That’s right. Thanks so much.
