In an era where automation and artificial intelligence (AI) are revolutionizing industries, wastewater management is (no longer) lagging. Kando, a wastewater intelligence company, is pioneering this shift with its AI assistant, STREAMi. This transformative tool aims to alleviate some of the sector’s most pressing issues, from workforce shortages to complex data analysis.
But STREAMi isn’t just another tech novelty—it’s a solution grounded in practical necessity and designed for real-world application. How? Let’s figure out:
with 🎙️ Ori Reshef – Chief Solutions Officer at Kando
Resources:
🔗 KANDO’s website
🔗 Ori Reshef’s Linkedin
🔗 WEFtec
🔗 KANDO’s first appearance on the podcast
is on Linkedin ➡️
Full Video:
Table of contents
Kando’s Quest for Quality
Water quality is an omnipresent concern that extends beyond mere compliance with regulatory standards. Ori Reshef, Chief Solutions Officer (CSO) at Kando, underscores that wastewater quality should be everyone’s business—from CEOs to the general public. With Kando’s advanced technology, utilities can now monitor and manage wastewater quality more effectively than ever.
Kando’s expertise lies at the intersection of advanced data analytics and environmental management. Utilizing contactless sensors and a sophisticated analytical server, Kando collects real-time data on wastewater conditions. This is where STREAMi, their AI assistant, comes into play.
Introducing STREAMi: AI in Action
STREAMi isn’t just an AI tool; it’s a game-changer in wastewater management (or, it will become it!). It leverages OpenAI’s capabilities coupled with Kando’s datasets and APIs to help utilities sift through complex data, providing actionable insights and even drafting detailed reports and presentations.
From identifying pollution sources to crafting action plans, STREAMi automates tasks that would otherwise require extensive human labor.
Imagine you’re a pretreatment manager grappling with multiple data streams and legal compliance requirements. STREAMi can monitor sensor data for you, highlight critical events like chemical spills, and recommend targeted actions—all while updating in real-time (still through the magic of Kando’s APIs).
What would have taken hours now takes minutes, freeing up professionals to focus on more critical problem-solving tasks.
Tackling the Workforce Issue
The water industry’s workforce is aging, and attracting fresh talent is a persistent challenge. Thirty percent of the workforce is expected to retire within the next five years, while only eight percent of new engineers are entering the field.
The sector needs solutions that can do more with less, and this is where STREAMi shines. By automating the mundane yet necessary tasks, STREAMi frees up human resources to tackle more meaningful work.
Additionally, Kando is keen on being a piece in the larger puzzle rather than attempting to become the overarching solution for all things wastewater. The aim is to specialize and excel in wastewater quality monitoring while leaving room for integration with other systems, thereby creating a more holistic approach to water management.
A Pragmatic Approach
One of STREAMi’s most compelling features is its adaptability. Whether you’re working under U.S. Environmental Protection Agency (EPA) regulations or adhering to local laws in Israel, STREAMi can tailor its advice and analytics to fit the unique demands of various jurisdictions. The API connections ensure that data remains secure and isolated, providing peace of mind for utilities concerned about data breaches and misuse. (a recurring fear in water digitization)
While the AI can dive deep into legal contexts and expert insights, it does so without overstepping its bounds. The system never acts autonomously but augments human decision-making, ensuring that operators remain in control. This harmonizes with the fundamental principle that AI should augment, not replace, human capabilities.
Building the Future
Kando’s journey isn’t just about deploying an innovative tool but understanding and addressing the core challenges faced by the water sector. From its inception, the company has emphasized the importance of high-quality data as the backbone for any effective AI solution. The focus remains on continuous improvement, leveraging the latest advancements in machine learning and analytics.
And what about STREAMi’s broader impact? Kando recognizes the potential for even greater integration and is floating the idea of a more accessible, light version of STREAMi available to a broader audience. Whether it’s utility operators, city officials, or concerned citizens, a streamlined, user-friendly version can help demystify wastewater data and heighten public awareness.
I’m told, we may be able to test it out for ourselves starting from the upcoming WEFtec 😉
Looking Ahead
Kando has set ambitious goals. By January 2025, the aim is for STREAMi to become the primary interface through which customers interact with Kando’s data analytics platform. In the pipeline are additional specializations of STREAMi tailored to different roles within the utility, from operational teams to public relations units.
This isn’t just about technology for technology’s sake. It’s about practical, impactful change. As Ori Reshef aptly puts it, quality should be the business of everyone. And with STREAMi, Kando is making this mission more attainable than ever.
Conclusion
Kando’s work exemplifies how AI can revolutionize complex fields like wastewater management. By addressing real-world challenges with innovative solutions, Kando and STREAMi are setting new benchmarks for the industry. As the sector grapples with workforce shortages and increasing environmental regulations, tools like STREAMi offer a glimpse into a future where technology and human expertise coalesce (nice word, I know, I verified, it exists!) for the greater good.
If you’re intrigued by what Kando and STREAMi have to offer, consider visiting their booth at WEFtec (and give me a poke). It might just be the glimpse into the future of water management you’ve been waiting for!
My Full Conversation with Ori Reshef on LLMs applied to Wastewater
These are computer-generated, so expect some typos 🙂
Antoine Walter: Hi Ori, welcome to the show.
Ori Reshef: Hello, Antoine. Thank you very much for having me. Very excited to be here today.
Antoine Walter: Well, i’m super excited as well because we actually changed One letter because last time I had Ari,
Ori Reshef: Yes, it’s true.
Antoine Walter: Ari was actually one of my first guests on the podcast. it was season one, roughly mid of season one, we had the conversation about Kando. there’s probably a lot we want to update, but.at the time I discussed with Ari, the reason for Kando to be in business was this element that whatever is happening in the sewer happens in the sewer and nobody sees it. there’s a lack of visibility about wastewater and you can make that smart.
Kando’s Mission and Evolution
Antoine Walter: Is that still the reason for Kando to be in business four years later, or what would be the number one challenge that you’re solving today?
Ori Reshef: That’s always our basis, we listen to the city from below. That’sone of our special sauces. What we see ourselves more and more, and we will talk about it, is ambassadors. Of quality of wastewater. We’ve seen wastewater a real resource. There’s so many countries in the world and cities in the world that are recycling, doing direct potable reuse, indirect potable reuse.
This is a resource and the quality of wastewater is in our gut. this is something we think that should be everybody’s business, your business, my business, the CEO’s business, the industries, the environments, the public. the first part, as I said, four years ago, this is how we started.
first of all, let’s know about the quality of wastewater. Let’s know about, what’s going on. And then we evolved into, let’s see what the source of the pollution is in an automatic manner. Let’s help the source behave differently. Let’s see the impact downstream and save the treatment plant from, a huge, stop in, activity, because the biological part was dead there was no oxygen and that could have been treated or worse, all the watersheds.
If we look around the world and see the public concern about the watersheds that we all go and with our families and kayak there and swim and everything. This should be guarded with a lot of attention. And so what we see ourselves, is, helping utilities first to cope with the wastewater quality challenge from one hand, understanding the upstream source and behavior and change it so we can improve the quality of the wastewater and making sure there is.
low to no impact on the downstream, which is the treatment plan and the environment. So we started off by seeing, analyzing and then helping our customers to do in action to change the behavior.
Challenges in Wastewater Management
Antoine Walter: But isn’t it a broad challenge here? When you say that you go from the watershed up to the plant, a ton of stuff is happening in between those points How do you focus?
Ori Reshef: As you said, we are situated in the middle. It’s a nice place to be. in the sewage network, which is exactly in the middle between the upstream, which is the city. The industry is life that produces the wastewater and the downstream, which is the treatment plant and the watersheds and the environment.
so quality as we see these everywhere, our first focus is to help pre treatment find the source. and change the behavior of the source so we can reduce the amount of pollution increase the quality sewage system and reduce the impact on the treatment plant.
That’s the first thing and what we see is that by doing that, our basic job, which is, helped understand what is the source of the pollution, change its behavior, alert and make sure that the people downstream in the treatment bed are completely aware of what’s coming in their way so they can prepare.
we understand that we have a lot of treasures that we cannot keep only to ourselves, that we need to, talk about inside the utility to the asset management people. For example, if you have breweries in your city, and they’re producing waste that, is very bad for concrete pipes, you want to know that and change.
So asset management, marketing, public relations, everybody is concerned by that. And what we’re doing is from our core business. which is, improving wastewater quality by finding the source and reducing the impact downstream. going more and more, and we’ll talk about it through GenAI, which helps us build quick solutions, but completely full based on our analytics.
Focus, and then spread the good news.
Antoine Walter: go to Jenny. I,because I think that’s the main topic for today, but before that, I’d like to understand where you’re coming from.
Ori’s Career Shift to Kando
Antoine Walter: you were having a good life and a great career up to last year. And I don’t get why you wanted to go into the poop side of the story, because you’re with SAP until July, 2023.
And I’m wondering how much of a jump it is to come out of a company like SAP and say, Oh,all of a sudden those things that were under our feet, which are just conveying poop, that’s going to be my bread and butter. so what’s your surprise when you jump ship?
Ori Reshef: all my life. I’ve been in large enterprises or the startups, but always around data, data analysis, data science, startup I was involved in was bought by, another, American startup did big data indexing, at SAP. I headed the product team for the customer data platform.
So it’s the unification of data, understanding insights and activation of, marketing automations to customers. my first love is data and my second love is nature. the planet. I always wanted to combine these two. I met Ari eight or nine years ago and we started talking I said, Oh my God, this is really interesting.
This is data that is so hard to mine. And to have and to understand, but it’s so rich. I wish I could get my hands on it. I spoke for a couple of years and I was consulting a bit as a friend because I was really excited about, the mission and really protecting people and the environment.
at one point I said, I’m old enough to say, I want to do something good not just for me, or the commercial side, but also something that will be of impact. A real impact that I can be proud of and tell my kids, listen, this is what dad is doing there for now, making jokes about the dad working in the poop business, , but they do appreciate it.
And I enjoy every moment of it. So the jump was. a very different business. What are utilities from, the big retailers or the banks, insurance companies, et cetera. you meet different people, but that’s the beauty of it. I met amazing people that are purpose driven and are there for a real reason, and they have this passion towards what they’re doing,
You cannot beat that. So I’m very happy with that.
Introducing STREAMi: The New Product
Antoine Walter: Let’s get to that Gen AI topic. I needed to have a special excuse to have Kando back and to investigate your path. And that special excuse came with a product you’re launching. I don’t know if it’s yet launched because I couldn’t play around with it,
Ori Reshef: True.
Antoine Walter: Streamy. And you know what, instead of me to explain what Streamy is, I’ll let you do the elevator pitch to Streamy and then we can discuss it.
Ori Reshef: Okay. So I’ll even do better. We’ll, have, our friends Streami. Join our conversation and let herintroduce herself. Hey Streamy Tell me what is it exactly that you do for a pretreatment manager, please?
How STREAMi Works
Ori Reshef: As a pretreatment manager, I help you by providing insights on upstream pollution sources And the overall health of your network. This includes one data monitoring. I ensure the quality of data from sensors in wastewater systems, helping to identify and prioritize pollution sources, such as factories or facilities, which can impact treatment efficiency to event alerts.
I highlight critical events like chemical spills or blockages that might cause overflows or system inefficiencies. Three action plans. I recommend targeted actions like increased sampling or infrastructure improvements to mitigate pollution risks. Let me know if you want details on any of these aspects.
No, just say hello to Antoine and the amazing podcast we’re in.
Hello Antoine to this amazing podcast. Glad to be part of it.
Antoine Walter: Glad you’re here, Streamy.
what is it? Is it like, , I can speak to chat GPT. I recall maybe six months ago, what did Corey did live where he was speaking water with chat GPT and you can’t speak water with chat GPT to a certain extent, it will simply not be as specific as Streamy just was.
So what is Streamy?
Ori Reshef: So, Streamy, sits under the shoulders of giants, let’s start with the basics.
The Role of GenAI in STREAMi
Ori Reshef: How do you create a stream? open AI application. Well, you have three ways of making the AI do a specific task and be an expert. this is part of the answer your question.
What is Streamy first of all? It’s an expert in wastewater. And, a waste water quality. It’s an expert in Kando. and how can you do that? There are three ways. The first way, is to train a LLM model. Not in our league. This is something that Microsoft will do that. Google, , Tesla, , Facebook, et cetera.
All the huge ones. They will have It costs tremendous amounts of money. Even in the American presidential, campaign. Now you hear debates on how much energy. And going into nuclear energy to power the training of a I model so can do. We think highly of ourselves, but we’re not that big. and
we cannot train the model. So that’s one way and not for a company our size. the second one is fine tuning. It’s not really building and modern. It’s taking a modern and, and training it on, on a certain percentage and making it, more adapted to your domain of expertise.
Also very expensive. We might get there, and, , and we’ll see how on the it done. By the way, alpha is already launched, two weeks ago and we have two customers already working with it, . Fine tuning will be the second way, but the third way, which is the quickest way, which I actually recommend to everyone, is what we call rack, which is basically, augmented research, generation, which is, Give context.
This is why when you talk to chat GPT normally it will be okay ish.It read Wikipedia, it read some articles in Water Online, etc, etc. But what does it know about a specific city in the U. S. and the industries and the risks and what is the chemical composition and the lab results? It doesn’t know anything, so I need to give it context.
context.what we did with our Streamy is basically we’re working on top of open AI. This is the engine we are using. We’re using our own API’s. we’re using the analytical layer. to remind everybody Kando has contactless sensors inside the sewage system that transmits in real time the data into an analytical server that has algorithmic pipeline that goes through and creates basically , those quality events, qualifies them, quantifies them, understands them, enriches them, and then arrives into insight.
Thank you. On top of that, all this server has API. So, we question with open AI, the KandoAPIs and server, and we give it , real time context. This is the situation right now. We give it all the legal, context of the EPA regulation, the permits, our own knowledge, Kando has, extensive knowledge of subject matter experts with years and years.
Are these one of them? Of course, our founder and Zohar, our co founder are real experts in wastewater. And we have a team, subject matter experts that can help us. Andtake their documentation, their knowledge and put it in instructions inside the system the end user should only ask a simple question and the simple question will get all this context and go to open AI and grab an answer that is with this context.
And then we leverage the two things what can do is doing all the analytical stuff and then open AI will give us. The answer, the formulation analytics, it’s amazing. I’ll show you some example. It’s mind blowing. I bet you played around quite extensively with these kinds of tools.
but our customers are now using it. And we’re very, very, very, attentive of data security and everything. So they’re discovering new stuff.
You’re asking questions I didn’t think of.
Antoine Walter: we’ll go to that because I’m interested in the why, but just for me to put a bit of context and to understand, beginning of this year, I was creating something which I call the H2 Oracle, which was basically, I took all the transcripts of all the people who went on the podcast.
So Ari’s insight was inside and I did a vectorization of that. And. It’s this rag element you mentioned. so used it, but I was also restricting chat GPT. So that’s chat GPT could not use its own knowledge. it could only use what was inside , my vector was pretty cool. got a lot of traction, got a good success at some point had to unplug it just because he’d got too much traction.
And these APIs costs so much that I couldn’t afford to have like, like, like a freebie with myWhich was costing me like 500 to 1000 per month in API. So, but what’s different in what you’re doing is that you have that layer of vectorization of the legal context and the expert insights.
But what you’re adding to that is that you’re connecting to the Kando API, which means that you have an updating and, live interaction with what’s in the real world. so it’s not like just my garage project you went. Above and beyond, because you have that API layer.
Ori Reshef: Yes.
Antoine Walter: Do you still enable ChatGPT to browse the web and to look for its internal knowledge?
Or do you restrict it to your vectors and APIs? what are the boundaries?
Ori Reshef: So I don’t like boundaries with this kind of tools because we are in exploratory mode and we don’t know what we don’t know. We’re talking with the open AI guys and then we say, Hey guys, we’re chatting with them because we’re developing and we need their support, working, not working, this API, et cetera.
And we’re asking them, Hey guys, can we do this? And they go, maybe. Test it and please tell us when you come back because we want to know.
Customer Feedback and Future Plans
Ori Reshef: The thing is this monster can do so much and what we did, we launched it with two of our customers. Now the third one is, next week we’ll go live. And what we told them, we told them the following.
This is an alpha product and we’ve chose , the more, advanced customers , that we have, please break it, please do anything you want, ask any questions. we added the feature, of thumbs up, thumbs down and comments. we record everything, every question, every answer, so we can analyze it later on.
We have, what we call inside the database, which is approachable with API’s, a risk assessment map, right? We know the entire geographical area with the coordinates of the industries with a lot of calculation on risks, etc.
And we didn’t put it in the spec of Streamy and it was not the intention. But one of our sales executive said, let me try something. Can you give me a risk assessment of 1, 2, 3, 4, 5? And it did. And it was perfectly good. the exciting thing about this is that we don’t know where the boundaries are, This is the time where we permit ourselves to go wild and go, as far and crazy as we can. But as you said, at the end of the day, our answers should be spot on. what you’re referring to is the very amusing, sometimes, effect of hallucinations of the model. Because it is trained to give an answer no matter what.
have a timeline, you need to spit out an answer. But this can be completely nonsensical, and we’re talking about a critical system, so we cannot say nonsensical stuff., short answer now, go as wild as you can. We’ll close it down at the end of the, pilot period. our customers are already going crazy about it because it really shortens the time
Antoine Walter: but let’s go into the why. Because there are two possible whys. And I hope for you that you’re for the second one, the first way is, you know, everybody’s speaking about, chat GPT and,Gemini and,whoever else you want in the word, because it’s not just open AI, but okay, it’s super trendy.
So we need to do something about it. That would be kind of a lame reason, but why not? And the second reason is that you identify something very clear where you say here, it’s going to bring tremendous values to our end users. I hope for you it’s the second one, but if it’s the second one, what is that specific stuff you wanted to achieve
Ori Reshef: so it’s the second one. And there are two problems that we wanted to solve that are interwind. because we believe quality is the most important thing that everybody should be concerned with. And we sometimes saw that, even CEOs and executives are talking, they understand the importance of quality, but they do tell you no.
Addressing Workforce Issues with Technology
Ori Reshef: I have this list of things to do and this list of people that should do it and everything is prioritized one of the major concerns of the water industry around the world is workforce I think the statistics is 30 percent are going on retirement in the next five years and 8 percent of youngsters are going in because you can imagine that young engineers finishing some kind of engineering school will probably go work for Google or Facebook or Amazon than their water utility.
, you need to have a good reason or a good internal motivation to say I don’t want to make this large amounts of money when I’m young, I want to go directly to what my heart is seeking, and it’s a hard domain. it’s not always a easy job. I think one of the more interesting jobs in the world, but not very attracting.
So we have a workforce issue. That’s one problem. it prevents our customers from doing the job. They need still to do their job and then job. Let’s not forget it. What the utilities give us clean water. And protect us and recover the wastewater and protect our health and the environment. Super important mission and they don’t have people.
They really don’t have people. They struggle with it. The second thing is is an analytical system. at the end of the day, we have a dashboard with graphs, pie charts, and, all kinds of insights that, a smart person should look. Analyze, understand, write an email, write a presentation, call a meeting, and then, in the meeting convince people that this is the right thing to do this is the data that supports it and let’s have a plan, et cetera, et cetera, until you go to the corrective action.
So you have a chain from insight to corrective action that takes a lot of time and resources, We don’t have people, we don’t have time and now We’re adding to the problem because we’re adding more information, more data that is important.
So you’re frustrated. You need to look at it. You need to analyze it. You need to take an action. You need to call a meeting, People are telling you, you know what? I don’t want to see that. I don’t want to know. you’re bothering me because you’re telling me stuff that will only frustrate me and I cannot do it.
So we said, okay, with technology, with Gen AI, you can have a little helper who’s smart, who knows exactly in real time what’s your status right now, and can do all those annoying tasks that take so much time for you. It can do the entire digest and ingest of the analytics, tell you what is the bottom line, tell you what you need to do, write the email for you, send the email write the presentation, send the presentation, follow the project, write the communication email, et cetera, et cetera, et cetera.
So there is, a real need for relief, relief in terms of people. The people needs a lot of minions all around them, smart minions that will just do the job,
Um,
Antoine Walter: and it’s alarming because I had that conversation with amanda sikora from vapar she was explaining how utilities would not jump boat because they don’t want the information. if they start having the information, they will need to act on it and they can’t. they’d rather, you know, what I don’t know can’t harm me.
So ignorance is bliss. Let’s not go into that. , that’s why it’s alarming that there is this tendency that we don’t want to look at information just because we don’t have the means to act on the information. to that extent, you created the augmented. Operator or the augmented executive or the augmented PR person or whatever, because you’re, aiming at solving the problem by saying we don’t have more people, we don’t have more resources, so how do we do more with less?
Well, we use tools. I’m with you on that. Now, when I started the podcast, one of the big questions I was asking around was, are we aiming towards. the automated network, the automated plants, the automated stuff like that. And so my vision at the time was we would not have something to augment us. We would have something to do all the boring stuff we don’t want to do.
And then the human becomes just the one to check what’s happening. you would not be the one acting and moving a valve. The system would act and move a valve and then tell you it has done it. It looks like we’re taking a different direction, which is to say.
We have to make it easier for the operator to take the decision, but still he’s in charge of the decision. So does that mean that you’re using all the intelligence you’re gathering with Kando to augment the operator, but you will always leave the operator in charge and never act instead of him?
Ori Reshef: there are points where you need human intelligence, I completely adhere to your vision. I think robots can go anywhere and automate the system and open valves and close valves, et cetera. But there is, a human factor that should be in the loop the human intelligence should be.
augmented. more and more automation should happen to make sure human operators are doing what they do best, with the mind blowing capabilities of GPT and all the LLM models. There are two things we can do. one is being creative and solving problems that are, more complex and don’t have a textbook solution and empathy.
Understand what the effect of a problem has on other human beings and take decisions according to these parameters. I don’t see LLMs going there anywhere soon lucky for us, we still have some advantage, but this can take us so far. one person can do much more and leverage their.
Human capabilities. much, much more, of the examples, when we started to work , with one of our customers, we did an investigation with Streamy, it gave us all the top polluters and showed some graphs at the end, I said, okay, Streamy, please prepare a presentation for the next meeting.
Streamy said, well, sure. Fun. Here’s your presentation. the room went wild. Why? Because they are engineers. specializing in numbers and understanding complex hydraulic problems and waste water problems. They hate writing presentations. And they told me, you know, at the end of the day, this small thing is hours and hours of suffering for me.
I hate it. I hate preparing the presentation. I hate presenting it. and I hate writing the email that surrounds this, you solved my problem. I said, Oh, this is the important thing. assembling the data, putting all the charts together, putting the bullets and we solve the problem.
Antoine Walter: Explain how you roll out. you said you have two customers, using the alpha right now. some questions on that. First, do you adapt the context to each customer? , let’s say One is in the US, the other is in Israel, and on one end it’s the US EPA, on the other oneit’s the Israeli National Water Law.
So you would not have the same legal context. I would imagine that it would eventually make sense to have for one, the US, and for the other, the Israeli regulation inside, and to not Cross the two. in order to not confuse people, so , that’s one thing second, you mentioned you picked people which are more innovation prone or with an open mind or creative people, which might be able to leverage that.
STREAMi Launch and Customer Reactions
Antoine Walter: What was the reaction when you brought Streamy and then I have more questions, but let’s start with those two ones,
Ori Reshef: first one, you’re completely right. we have two customers. the three that are doing two launch in the US and one is launching next week in Israel. context is different. when you look at the U S as a target market for a company like ourselves, it is very, very different.
water scarcity is different from one state to another. The water laws are different from the way people look, the consciousness. And so we must have different, contexts. per client, especially in the US where you have the permits and it’s per industry per watershed capacity per the state of mind of the legislature later in the specific state or county.
Antoine Walter: So . That’s an advantage. That we can give context the context is per state and sometimes per county and sometimes per region. So the context is specific to each one. The API, would it connect? only to the data of one customer, or would it be like a worldwide anonymized stuff, which you can then leverage?
Data Security and Segregation
Ori Reshef: And we’re very, very, very, very careful with that. Data is completely segregated. at this point, we’re not even dreaming of doing aggregated. It’s in the vision, but the first thing is data security, for our customers. So data is segregated. APIs are completely tokenized andsecuritized andencrypted
one customer can see only their data in their context, which helps us also to be better. Actually, one of the nice things is that, when you start the thread with, chat GPT, it holds the memory of the questions, and the more context it has, the smarter it gets and more surprising insight you’ll
Antoine Walter: how did they react to Streamy?
Ori Reshef: The reaction was, wow. Because you can talk to Streamy like your smart friend and they really started talking. I think the major concern is always the same concern. It’s the right concern. This is our largest battle.
Challenges and Concerns with STREAMi
Ori Reshef: And this is something that we should talk about and how we deal with that, which is data quality.
Everything at the end of the day is about how good is your data. And how clean it is, how reliable it is, how accessible it is, how it is organized. This is the basic of everything. we can do no magic on top of it, but if your data is garbage, it’s garbage in garbage out And that was always the concern because it kind of augments the concern
Cause you’re talking to a friend and he’s very convincing and he’s very smart. Telling you nonsense and the risk is to follow nonsense. So their major concern, once they said, okay, this is cool, this is working. But is it real? Is it really real? this is what we’re doing.
We’re making sure that every answer is real, we have, developers. We have code writers, that are software engineers, and we have, a team of researchers, divided into experts in wastewater and data scientists. something that is always on and will always continue to invest resources is the quality of the data that we output and the quality of our analytics, the machine learning models that are producing the, events.
Always, a roadmap. there are always, two data scientists that are on that. They don’t see anything else. They don’t know anything else. They know on how good is our data and make sure it’s always better. That’s the concerns of our customers also.
Marketing Strategies and Product Ideas
Antoine Walter: What’s a major perk of Streamy could also have a downside because if it’s highly customized and really tailored for everyone’s use. That means before every deployment, you need to adapt Streamy to make it tailored to something. I was thinking, . you know, WordWave is a company who launched a mini product called Twitter Roast. I’m not, So much of a Twitter guy. So it doesn’t speak to me, but I saw all the Twitter guys starting to say, look, I just plugged that mini product on my Twitter. It looks at all my tweets and then it roasts my Twitter , and it’s super accurate.
And it’s a product they don’t sell. It’s a product that really it’s offered for free. And it has this kind of viral capacity because you want to test it because it’s cool, and you want to share the results. everybody gets to hear about. The word wave company, because there’s , this mini product.
And I was just looking up, what are the other companies doing that? And you, you have copy AI is doing a cold image generator. Wix has a QR code generator. Ahrefs has an SEO keyword generator. And so far and so on. The one which I’m using is co schedule has an headline analyzer. Before I put a title to my podcast, I go to the headline analyzer from Kosky.
I wouldn’t even know that Kosky exists if not for that. And so I was just thinking, you know, is there any chance for you? Because one of your hats is that you’re also in charge of marketing and product marketing. Would it make sense for you to do, a streamy light, which doesn’t connect to the API?
Because that’s the kind of, analytical layer, which you don’t want to leak out because of safety concerns makes a ton of sense, but has all these legal context expert insights you’ve collected and hence , is a cool tool that’s, you know, stupid people like me will like to record themselves chatting with that Streamy-Lite for three minutes and then share it around LinkedIn.
And so you create. Virality and you raise awareness of Kando, hence you use it as a Trojan horse. Does that make any sense or it’s really not what you’ve intended it to be.
Future Vision and Product Development
Ori Reshef: it’s a bit scary. It’s as if you were here today in my office, listening to our talks, because this is exactly what we’re thinking. 200 percent agree, and the idea ofa product led growth approach, PLG approach, with that is amazing. As you said, it’s always a tricky part, PLG, when you do it a couple of times, how do you create the connection between the free version and the paid version, but I’m putting that aside.
It is viral. What we’re looking for is exactly what you said. We’re looking do a kind of autonomous Streamy that adds some kind of value. We have a couple of ideas to embed it, if you’re the mayor of a city, you can embed it in your website and people can chat about their risks as well.
and for example, for, for the entire us we have a map for the UK and Ireland. We have a map, Europe, some of it we have a map. So this is something that we are thinking of completely in the marketing,side. and I invite everybody to Weftec to see , what will cook up until then.
hopefully you’ll be there and we can chat, in our booth over a good, coffee or beer.
Antoine Walter: So WEFTEC is the place where Streamylite comes to life and then you create virality around it. And that’s a clever marketing move. be the contrarian now for a second, because all of that is super cool and I’m pumped up. I want to also take the other side, which is, get your, your,explanation as to how you can augment.
and expand the capabilities of the water professionals, the engineers, the executives and everything on the other end of that same story, everybody is screaming for their attention. Like now your sewer could talk to you, but also your pump could talk to you, but also your, SCADA could talk to you, but also your, GIS could talk to you and so far and so on.
Everybody wants to be like your best augmented friend. So you’re fighting for their attention You know, I’m trying to connect that to your previous employer. And
Ori Reshef: Hmm.
Antoine Walter: the ERP world, there’s SAP, there’s Oracle, and then there are some small people, but it’s like everything is in the one, one roof, whereas in the water world, I could just look at the alumni of my podcast find 20 or 30 different people who want to be kind of the water OS and , to be , that full thing. Long question, which leads to a very simple one at the end. Do you intend to build like the wastewater slash sewer OS, which covers everything or in the long run, will you be A puzzle piece, which vertically integrates into whoever becomes that OS, you are just a link in the chain.
Ori Reshef: you’re right to mention my previous employer at SAPI I was the os. SAP is an amazing company. almost all the Fortune 500 companies in the world use SAP and you can be the OS of the entire enterprise. there I learned how difficult and how much responsibility and how big of a company you need to be to be the os.
For us to think about that at this point in our lives is not smart. we need to be the best in the world in wastewater quality analysis, finding the source. And, reducing the impact on, on the environment, I think it’s closer to a piece in the puzzle that can easily go anywhere you want.
I think the secret and you’re very right. is the explosion of data. When you talk with, with water, people say, not that I don’t have too much data. As you said, I don’t want to see it anymore. You’re telling me too much. And you’re kind of painting the image, which is very similar to that.
And I think you’re right. Everybody would want to talk to you to have your chat assistant, to have your augmented, whatever. I think the secret will be. To be very, tailored to solve a problem for a person in the most complete way, but one, not the entire thing and be where they’re used to work. meaning that if I have my phone with me all the time and I’m used to working with my phone, why should I ask you to open a laptop in the field?
if you’re already used to it. this is part where AI should be seamless and Gen AI should be seamless. I want to see it in my Outlook. I don’t want to have another application that is analyzing my emails and writing and suggesting and blah, blah. I want it to be to be done in Outlook, or in WhatsApp.
I want a chat bot assistant in WhatsApp analyzing everything. Tell me, do you want to answer this? Click here. I don’t want another app. So someone will come along, maybe one day and unify the, water world. there are a lot of startups that say we will be the water OS.
was this, small company called SAP to be the OS. Oh my God, you need infrastructure data infrastructure. You need, a lot of resources. It’s a huge headache. Just think about your data model. Is it covering everything? You know, all your sources, there are so many engineering questions that are so hard.
I think it should arrive, but it should be done the monsters of the market, and they should pick and choose experts the domain, we want to shine. in our domain. we get this question actually quite a lot saying, given that you’re so data oriented and you know your stuff in data and you have your infrastructure, why can’t I pour everything that I have to yours, server?
You do your magic and spit out something . That’s a much, much larger company with a lot of more investment and lot of lead time until you can do something as complex as that. So we’ll stay humble and pointed.
Investment and Profitability Goals
Antoine Walter: That’s a perfect transition because when you mentioned big companies and how big of a company, I was just looking up, the investment path of Kando and Kando since Ari was on that microphone, raised an 8 million Series B in 2021, which is the last raise you did, which in That world means whether you’re incredibly profitable and you don’t need to raise, good on you, congrats, or you’re bootstrapped and you have a lot of people working with great passion to develop it and you don’t need to burn money, or you’re about to raise.
So which one is it?
Ori Reshef: So we have an open round that we almost. closed our existing investors really believe in us and , they increased , their investment. we’re not profitable yet, but this is our aim. very shortly. so we didn’t do a big raise. We did, a smaller raise that will enable us to, go and explode in the market with what we have.
Antoine Walter: in general, I think profitability should be the North star of everyone. that’s the hardcore truth, right? , let’s go back 50 years ago, there was no question, growth or, profitability. You need to be profitable in order to be a business. think we’re going back to this very sane logic So in the context of a Streamy, is Streamy meant to be profitable or is it not? One step in the process and then the overall Kando experience is meant to be profitable.
Ori Reshef: The overall Kando experience is meant to be profitable. Streamy, is a part of Kando, doing all the things , to make our users life easier, quicker and really gain , the riches , of Kando, but it’s,, a part , of what we do and we clear upstream. We make sure your upstream is clear.
now we want to do it in your context and augment, our users. But it’s the Kando experience , vision in terms of generative AI is again build specialized solution for specific roles in the utility, learn them, know them very well, learn, understand what they need.
And give them this solution and with this we can augment our user base and the number and types of people that are using, Kando and then again, going back to our mission, quality is our mission. And since we think quality is the business of everyone, everybody should use their own Streamy.
from their perspective. If the perspective is public trust, if the public does not trust you, you have a problem as a utility, you have a problem as the mayor. But if you have a stream that is always there saying, Oh, look, improved by two quality points, the wastewater quality, and this is the impact of it.
I know it. I stream in. There you go. There’s your, LinkedIn posts or Facebook posts or your press release. Just click here. if the CEO sees their CSAT score, NPS score goes up. Then they said, okay, I have a use for this small gadget that I can do put there.
Antoine Walter: So Streamy is your UI . But the breadwinner is ClearWebStream, which is actively bringing all that data, collecting it, putting it together in an API form, and then Streamy. Enables you to tap into the API
Ori Reshef: it’s so important what you just said, because people tend to mix everything up and say, LLM is magic. And there is the hard statistical mathematical work and hard hardware work. that needs to be done. And this is why we do not give up. On the contrary, double down on the machine learning model, on the anomaly detection model, on the quality of our hardware, on the accuracy of the hardware.
Streamy is an amazing addition. It adds a layer of intelligence and analytics, and you can ask for any graph that you want, which is amazing. But. It’s a conversational UI, very smart but it sits on the shoulders of giants. giant is what we do our core business and our core business is to know waste water quality and to improve it.
if you don’t have that, then okay, you can go and to chat GPT lambda, and that’s the question and get the answer. And that would be fun, but it wouldn’t get the job that
Antoine Walter: and you’re not a super big company, but you’re still orders of magnitude larger than me. So you can support the API cost and it’s okay for you to keep building it. I get it.
Ori Reshef: I admire the attempt because it goes up really quickly.
Antoine Walter: Yeah. Yeah. I can talk to you much. it was expensive, but yeah. At least I learned something. you mentioned you’re in the closing of your financing rounds for people listening to us, are you. full done and just closing, or are you still taking calls?
Ori Reshef: we always like to talk to, investors that are really interested in that.We’re looking for partners. We’re almost done. So, not much room left, but if you’re interested, contact us. we’re always happy to talk. We learn from every interaction.
So always happy to talk
Antoine Walter: And to close that deep dive, do you have a KPI for Streamy to say by the mid of 2025, you’ll have rolled out to 20 customers What’s a realistic target for you?
Ori Reshef: the roadmap, we did it very quickly and I have to mention the super amazing team. we’re a 50 person, company. we have super talented, and motivated people. We have, Nadav Karpenkov, who’s our VP R& D, who led this. Or Nisa, our VP product did an amazing job.
The entire R& D team, Dikla Raz from the, our head of research. we invest in those talents and they surprised me by the speed and quality of work we were able in a very short time to go on production and now working with the customers. So we became more ambitious we want Streamy to be the way that our customers interact with Kando, hopefully , by January 2025, it will be, there will always be the dashboard and you can always fall back to what, you know, But , that’s our aim.
another aim we have is to produce two more Streamys. One, as we mentioned, the more, open Streamy , closed environment that everybody can play around with and have fun. , and getting more and more we have, in our roadmap and operational stream for more operational people that will help you optimize around on the network and, and, and manage it, the PR system,
we have lots of dreams and we’re executing, so it’s fun.
Antoine Walter: So January, 2025, you go out of alpha, you’re going to beta or live.
Ori Reshef: Yes.
Antoine Walter: didn’t hear your targets. How many people, how many customers should have it by the end of 2025?
Ori Reshef: At least 20 customers should have it.
Antoine Walter: At least 20. it’s a bit frustrating because I don’t want to take you for one more hour, but we have more topics because I’d like to understand at some point on the line, the focus of Kando on the people, the focus of Kando on the workers, and also the focus of Kando on the people.
PR communication side, because those are two directions, which are pretty unexpected from a water company like yours, but that’s great because it will build the excitation to be around that WEFTEC and to have that talk and the beer and the drink and whatever it is in the meantime, thanks a lot for having been with me on that deep dive and that LLM route, if that’s fine for you, I propose to switch to the rapid fire questions.
Rapid fire questions:
Antoine Walter: First question, what is the toughest challenge in your opinion for a water tech startup?
Ori Reshef: the fact that they are not working in front of technology companies, the technology is not the main business of. What are utilities and they need to understand to explain the technology in terms of how does it help me in life? And so I urge them to go and understand the problem that they’re solving and not fall in love with the technology, which we often have.
Antoine Walter: What would be your best single piece of advice for the founders and managers of the about 1, 000 early stage water startups?
Ori Reshef: Study the problem. study the market and make sure that if you’re, leaning on data that you have it And if it’s real time that you have a way of acquiring it I saw, unfortunately, startups that developed amazing stuff, but the lead time to the data or the status of the data was, A barrier
Antoine Walter: What’s the drop of knowledge you wish more investors knew about the water sector?
Ori Reshef: water should be looked at and waste water should be looked at as one of our most important resources. It’s like gold and like oil and like gas. You should look at it like that and you should look at it long term. And you should also look at climate change and what does it mean to water scarcity and water availability.
What was your most unexpected partnership and what did it bring you? in can do in a professional life. I think the most unexpected partnership was between the two poles of product house and, customer experts this partnership, this friendship between the two parts really improved our product understanding and, and,and roadmap in ways I cannot imagine.
it’s not. One against the other, it’s really a strong partnership. The product is everybody’s, especially those who sits with the customer. So partner up with them, talk to them.
Antoine Walter: profitability or growth.
Ori Reshef: Profitability,
Antoine Walter: What’s the next profile you’ll hire
Ori Reshef: data engineers,
Antoine Walter: and when you hire are looking for sector experience or startup experience
Ori Reshef: sector experience,
Antoine Walter: opening your markets or doubling down on the current ones
Ori Reshef: double down on current ones.
Antoine Walter: . What’s that tool nobody speaks about, but you couldn’t live without.
Ori Reshef: have to examine one that everybody’s speaking about. I’m addicted to chat, GPT and Gemini. I mean, both. I’m using both for everything in my life. It is embarrassing to say how many domains are now completely there. I pay both, open AI and Gemini, so I’m working a lot with them. and the other tool is an electric pump from my motorcycle.
That’s I cannot live without it.
Antoine Walter: What’s the single piece of insight your ideal customer profile needs to hear right now.
Ori Reshef: What is the public trust? Regarding their work.
Antoine Walter: What are you desperately needing and want to raise an open call for right now?
People that want to try out and play around with us, come to our booth, come to Weftech, check out Streamy, check out what it ignites in you. We’re looking for people that have an open mind and a passion to explore. So the way to try out Streamy is to come visit you at WEFTEC at your booth.
Ori Reshef: Yes, definitely. And for you, especially, you’ll get, a URL, once we’re building a separate environment that is separate from our customers, you’ll be the first one to use it.
Antoine Walter: it. What can and should I do for you?
Ori Reshef: play around with Streamy and tell us honestly what you think that will be the best thing.
Antoine Walter: I’ll do that. That’s within my reach.
Ori Reshef: Perfect.
Antoine Walter: Thanks a lot, Ori.
Antoine Walter: If people want to connect with you and follow up after that conversation, where shall I redirect them the best?
Ori Reshef: either LinkedIn, you can find my profile or ori.reshef@kando.eco. Email me. I’ll be more than happy to answer.
Antoine Walter: As always, the links to your mail and LinkedIn are in the show notes. if you’re listening or watching have a look. I think we’ll have a good chat in New Orleans. So I’m looking forward to that. Thanks a lot for having been with me and talk to you soon.
Ori Reshef: Thank you again for having me. It was a pleasure. Thank you.