DTM E43. Future of Recycling using AI and Robotics - Ishitva Robotics
From the Deep Tech Musings podcast - Get first hand insights on how to go from idea to traction in the deep tech space.
Jitesh is the founder and CTO of Ishitva Robotic Systems, which is building futuristic automated Material Recovery Facilities (MRF) that can sort different types of waste at high volume for Indian Cities. Automated MRF solutions incorporate AI Vision Systems, Mechanical Robots, High-Speed Sorting, and Live Analytics of Waste Streams which Jitesh has developed at Ishitva over the last 6 years. Before Ishitva, Jitesh had 12+ years of experience in the software industry and worked for Fortune 500 Companies building products and solutions across the globe.
In this episode, we talk about how Ishitva can act as a catalyst for the “waste circular economy” by providing a high volume of materials to recyclers. Jitesh highlights how to evaluate the technical viability of AI/ML solutions and how to make clients trust AI. He further ruminates about the prospects of making his vision system available as a service. We end up with Jitesh’s advice on making it from idea to traction in deep tech and how the technology space in Gujarat is moving ahead in Automation R&D.
Listen now on - Spotify, Apple, Google
Show Notes
[00:01:02] What inspired Jitesh to get into entrepreneurship?
[00:05:20] What is the problem Ishitva solving? What is the circular economy?
[00:08:24] Ishitva’s traction till now and major clients
[00:10:05] Ishitva’s hardware and software technology stack
[00:15:02] How to assess the technical viability of an AI/ML solution early on?
[00:20:43] How to make consumers or clients accept and trust an AI product?
[00:23:09] How can deep tech startups create competitive moats?
[00:28:44] Like large language models, can Ishitva’s vision system be offered as a service as well?
[00:31:00] What are Jitesh's top suggestions for launching a deep-tech startup or product?
[00:33:48] Jitesh's viewpoint on the developments taking place in Gujarat in the technology domain
Where to find us?
Jitesh Dadlani – jiteshdadlani (LinkedIn)
Ishitva Robotic Systems – Ishitvaroboticsystems (LinkedIn), @IshitvaSystems (Twitter), ishitva.in (Web)
Pronojit, DTM Podcast - pronojitsaha (LinkedIn), @pronojits (Twitter)
Transcript (generated by AI, so please bear some typos and malformed words)
PRONOJIT
Hi, everyone. I am Pranajit. And welcome to another episode of Deep Tech Musics. I'm glad to have with us today Jitesh, who is the founder CTO of Ishidwa Robotics Systems, which is solving real world problems of waste by building solutions based on industry four pano tools and that includes the likes of AI, ML and IoT. So Jade, welcome to the show. How are you doing?
JITESH
Thank you. I'm doing great. I'm doing good. And thanks for welcoming us and giving us an opportunity on your podcast.
PRONOJIT
It was fascinating when you showed me that virtual video tour of your factory in Amdabad just earlier in the day and so I've all pepped up to learn more about the shitwa and walk us through our discussion.
JITESH
Yes. So that is our new innovation center. We came up with 2022. So after the COVID So since February we have started this new center where we do our RND, our demo center and our production in Amdabad.
PRONOJIT
Got it? So I want to step back a bit and I want to start with your background and journey. I know you mentioned that you have been a professional in the industries across geography and then you took the jump into entrepreneurship. So what motivated you to get into entrepreneurship?
JITESH
Okay, so a little bit about the background. I am a BAC in Electronics and Physics. Then I did my MCA through amdagh. I went to Bangladesh as a normal software engineer. I started with Mine Tree Planet Asia and then went to Dubai for three years handling CRM. I rolled out CRM for ten branches in UAE. I got an opportunity back in Putney where I did MetLife project for US migration project for IBM. I was a Java Jwcuting by that time. Then I came into this service oriented architecture. I got an opportunity with TCS where basically there was SOA architecture last lead was my data matrix thing where for McGraw Hill, a production house in UK. I had an opportunity to work on portals basically on the cloud. So I have a different domain experience where product building was the core like from business analyst role to designing to technical architect and being a software developer. So I was lucky and privileged to get all these experiences. Also, when it being a 15 to 16 years of journey, I always believe it is a product part which I was inclined into, not the services part or just solving plain vanilla bugs every developer wishes to be in a product part. When I came back to Amdavad, I wanted to use all my knowledge which I have gained and I was inclined towards IoT. So every Saturday, Sunday I used to work on this is six years back. Seven years back I made an Arduino raspberry pie. When they came, I started building an embedded team. We made smart bins. 50 smart bins. Waste was opportunity where no one wanted to enter. So I made Smart Bins for collecting a waste. We deployed 50 Smart Bins which had ads revenue on them. Fortunately, or unfortunately it didn't work out. No one wished to give an ad on a Smart Bin. So we miserably failed over there. I would say later we found out the point of collection of waste. The problem was sorting of waste. So sorting of waste was the problem because it had mass volumes. Because at that time, 2016, there was a study done of search Barad mission saying that 4000 towns were studied and I think every town generated more than 400 to finite metric tons of waste. This was 2060 studied. So I saw this as a bigger opportunity where, you know, humans were involved to sort waste. So my so my inclination was to solve a bigger problem. Coming back to the problem, when I went into the field of waste management I spent two years into a waste management company. I saw ten to 14 labor sorting one ton of waste. So just imagine a city like Amdaba generating 4000 metric tons of waste per day today in 2022, how many labor it would require. I saw automation in other fields such as car manufacturing where the painting was done through EVP or homegrown robots. I also saw automation in Flipkart and Amazon where the parcels were sorted and you could now easily see automation robotics across all the industries. But there is no automation, unfortunately, in the waste sector, which is an untapped domain. So hence this motivated me to enter into this domain which was totally untapped. And I was excited to solve this problem by building a material recovery facility across India, at least 4000 towns soon.
PRONOJIT
Great to learn how you saw that opportunity of applying some advanced technology to an industry which has a lot of manual labor and is not keen on innovation. So can you help us visualize what are these wastes like?
JITESH
Yeah, so I'll explain you through the problems. Basically, the waste industry, the waste starts in the morning, let's say from the kitchen, from serving a tea, let's say our parents or mom or any other using the milk pouches or the milk tetra packs. It goes to a bin and then comes the other stuff from office waste such as paper, cardboard or organics or vegetables or plastic shampoo bottles or a Stepler pen or a glass or everything. So these are the different types of waste. So waste can be categorized into paper, plastics, metals, e, waste, biomedical, glass, rubber. So they are all finite resources on earth, as we all have learned in our science subjects earlier. Now these waste are smartly in India, if we all understand, like our Pelari, various or the aggregators who come to our homes, they pick up smartly the paper and cardboard and we sell them the milk pouches to them and the plastic per kilo our moms used to do before the valley or every monthly. And they aggregate the waste. And the paper industry, they take that paper and the pulp is made, the cardboard is made, the plastics are again shredded and again a new plastic is made. Similarly, glass has been broken and again a glass is made. Metals are also melted and again nickel to nickel, copper to copper. So I saw that opportunity and it is currently segregated through hands. In India there are different types of streams of waste. One is a municipal fart waste which has 50% of organics and it hardly has seven to 8% of recyclables. The recyclables means which are finite resources which can be recycled again and again for the circular economy. Now, what is this terminology, circular economy? See if a resource is made, let's say a pet bottle, such as a biscuitry or a Coke. Once the single use is done, it goes to the bin. We can bring back, make a pellet and again it can be made polyester, or it can be again converted to a cloth or anything else, or a bottle to bottle. So the future is 2025, as crude is also finite resource. The byproduct of crude is plastic. So everyone wants to use the finite resources efficiently and that is what circular economy says. So our startup is basically a catalyst to this circular economy by bringing back that material to the production house, again, by collecting the material sorry, the sorting the material through this municipal waste or the aggregators. And there are these different aggregators with different streams. It could be a paper aggregator, a plastic aggregator, a metal aggregator or a glass aggregator.
PRONOJIT
Got it. So who are our major clients till now?
JITESH
Okay, so we are coming up with the biggest pet aggregators of India. We have also deployed to two of the MRF solution in India. It's been two and a half years where our sorting solutions are working to an MRF plant, the pet aggregator which we are going to work, which I can't name currently, but we are giving them a complete automated plant. Without human where 16 to 17 categories of plastic would fall in a compartment. Or they call it as silos and then it can be shredded also automatically, so that it's like a just in time period of data where in a single flow nothing, no time is wasted from the tipping area to the output part. So this all management theories and lean management and all those things we're trying to bring up through automation and industry for this waste domain.
PRONOJIT
Got it. Sounds great. I believe your entire team is in Amdabh.
JITESH
Yes, you're right. So we have different departments. We have software department. Again, it can be broken down into two parts AI, ML and other programming language. We also have a cloud like main one department which you are making for the portal part. Then comes a mechanical department where we have designers and we have production. We are coming up with a new team of commissioning and erection because many times you have to deploy the solution. So we are building a new senior guys for the commissioning and erection. Apart from it, yes, we do have the HR teams, the sales and marketing which we are growing with seniors. So today we are a team of around 35 to 50 people based out in amber. We are growing to grow by 100 engineers by next year.
PRONOJIT
Got it always good to see that growth figure coming in. And this brings us to an important discussion point as well, as you mentioned. So what are the major technologies at play in our product?
JITESH
So, the major technologies as we are a mix of different science or different engineering, mechanical, Plc, SCADA, automation and software. So I'll start with the software part. So from the software part we do use the basics of CC Plus. We also use lot of computer vision algorithms which are again written in CC Plus. We do not use more of Python and all those things because we are more with the hardware part. So we have to use lower level languages on the cloud front. On the portal side we use lot of means, tab kind of thing, mongo, Express, Angular and Node. We were given credits by both Google and Amazon. So we are thankful to them. So that is one thing. Now comes the part of the mechanical designing part. So we use the standard CADCAM software which are available in the market license. Once they do our two D and three D modeling nowadays they also do offer the AR VR which is very good to showcase to the client before the plant is manufactured. You can see the whole plant actually in 3D AR and VR. So this is what we are also trying to give the client a new better experience.
PRONOJIT
Okay, yeah.
JITESH
So going further, we are coming up with this data analytics part where a lot of data is generated of the material recovery facility for Per City. So everything is connected to the cloud and due to we have lot of sensors so everything the IoT, all the IoT says they send their analog and digital through our big data that is MongoDB. And on the cloud we do a lot of analytics and we present it. Currently it is one way, so we present it to the customer so that when he wakes up, he doesn't have to run to the plant. We want to give him a comfort zone. They should see the plant running. Who is executing, what is the volume passing through the material recovery facility? What is the data saying? Is there an issue in the motors? Is there a issue in the temperature? How are the conveyors working, how is robotics working? So every data you can see on the dashboards with the volume also we are coming up with futuristic MRF, that is the metal recovery facility or the waste auto plant which will see them today. This is the revenue generated, okay? This is a space which you will utilize to sort and after that where it will be stored. And they would be connected to the CRM and the ERP systems.
PRONOJIT
Great to see that end to end vision of the platform. And I think often those basic visibility across the chain is something which adds more value than a high end solution or some other advanced technology. But you wanted to mention something?
JITESH
Yeah, I want to add one thing. We are also plumbing up with brand sorting. So let's just sorting for a particular group of brands saying Unilever or PepsiCo. They are sorting their own items and we can give like this city or this MRF has generated this many visitor bottles. So you can give a sales outcome. Also saying the analytics, okay, these are the sorted items, these are the Pet bottles, these are the medical ones, these are the food category waste, this is a non food category waste, this is an oil category waste such as Castrol or a particular brand. So that helps for an EPR, that is an extended producer responsibility which is coming through the government of India where every plant manufacturer has to collect waste. So that analytics also will help them to collect waste also for the brand owners.
PRONOJIT
Got it? So I'm very interested to know that while most of the industry is using Python and other languages and here you are solving a real world use case at a large scale and using C and C plus plus. So any nuances there? Why you choose that?
JITESH
It's simple. Why? Because we have to work with the hardware. We have to use the maximum throughput of the hardware through thread processes and OS. So we have a custom OS, we have custom threads as per we want to use the full power of the processes. We use a lot of neural networks, we use lot of processing. Once you have to connect to the Plc, discard the automation part. So you have to work with the lower protocols. You can't work on Http or Https so protocols such as Mqdt protocols, such as TCP protocols, lower than that thing mode does also. So when you work on those protocols, especially on the plan, you have to use lower level languages.
PRONOJIT
Got it? And with AI solutions, there's always a challenge of how you ascertain the technical feasibility of this solution early on when you have very less data and you don't have the right specs to understand. So how did you tackle that challenge?
JITESH
So the first thing is to understand the problem. So for that thing I spent two years in a waste management plant. I saw how the workers were working. I went to a plastic sorting guy. So what they used to do is there are some eight categories of polymer. So they used to pick the plastic ones. First the transparent one, then the SGP, that is our hard pick bottles, high density polymer. Then our polypropyle trays that are recyclable the somato ones, lunch and dinner they give to you to eat your lunch or dinner. They are also been recycled. And then comes the shampoo bottles and all those things. There is toys, PVC items, electronics. So what I understood is how they were sorting through hands. So you have to basically mimic what the humans are doing in sorting and you just have to be a little high accurate and give a high volume. That is the first solution for the problem. You have to give not to do that thing. The humans were let's say they were how were they selecting it? So they were seeing the brand, they were seeing the texture, they used to push it, they used to press into their hand. So we learnt if we could store all this information in different classifiers, let's say in detection or classification, then the first thing was to understand are we able to make this hierarchy, are we able to store all these brands together and make a proper classification of different polymers? In waste especially, the challenge was the bottles were half or sometimes the bottles had dirt on it. So it is India. So the waste would be sticky, it would be grossy with oil and dirt and it could be brown. So how do you learn that thing? See, when you do an AI on a cakeLe or everything, you get clear data or you might have to clean the data. But waste is in India is totally dirty. I usually use three words it is dirty, it is dull and it is dangerous. So we had to basically we use computer vision a lot to basically classify them. We started manual labeling earlier and then we came up with all this new technology of synthetic data and all those things and today we are getting more than 95% of accuracy in detection.
PRONOJIT
Awesome.
JITESH
Yeah. There is one more important point. With the AI also comes a mechanical part where the robot has to pick in place or the pneumatics has to fire or eject. It's not the humans because humans use their hand which are connected by their brain. So I had to connect this AI part to the mechanical part and it also had to have that speed and that is where this lower level programming helps you out. Because the right hand side part of mechanical unit may only work with a good C language or an embedded part. Correct. So I would always suggest like, if you're working with hardware, you better go with C or embedded C or something like that.
PRONOJIT
That's a very great advice to act upon. And I also like I think what you mentioned is just like product development, one on wherein you get in front of the users or the processes that you're trying to improve and first understand that before even applying any technology or any before slapping a solution onto it.
JITESH
Yeah, it should be like the problem should be understood and you spend more time with the problem. We always believe that thing. The more you spend time with the problem, the more you understand it. What the user wants, what the community wants, or do they really require solution? And how scalable is it? So if the problem is scalable, if the problem is genuine and they do not have a solution if they have a solution I would request please do not reinvent the way if there is a market available solution and if either you are cost effective either you have a better features there has to be some add on value I would say to the customer then only you should find out in the existing environment do they have a solution? What is it? So in my case, the way stomach didn't have a solution, like, no one wished to give them an automation solution. So maybe that untapped thing helped me. So I sat with them for two years, I understood them, I went bit by bit. Let's say I started with a biscuit bottle sorting, then I just sorted Pet, then I sorted the second polymer, then the third polymer. Today we are able to sort paper, cardboard, multilad packaging, such as Wafers. We are also able to sort in multilayered packaging like packaging which has aluminum layer, which doesn't have aluminum layer, things like that. So waste will come with all sort of things. Today we also do color sorting, hybrid sorting, like mix of both glasses and colors and everything.
PRONOJIT
Got it? And it's important to know the time frame you mentioned that you spent in this problem discovery that's two years. And many of us look impatient as we are today. So we just want to get a problem and just slap a solution, whatever is available out of the box and do something. But understanding that problem as you sit through those two years to fully know the ins and outs and then trying to solve it, I think that really stands out.
JITESH
Yes.
PRONOJIT
Okay, so next I also want to have your view on another challenge which folks in the AI community generally face is they know a problem, they have a solution around it. But how do you make your users or client trust the AI? Did you face this problem and how did you go about it?
JITESH
I think I did a reverse thing I always started with a problem. I had a problem. I had a genuine problem where I spent time. What I understood is spending time with the actual users or the customers and understanding their business problems. How are they earning, what are the lagging? For example, in my case, the devaste managers, the devicete management companies, the manual ones, they were only able to do three tons of sorting through laborers. So once I give them our air sorters, let's say per hour, they can sort three to four tons with a two meter wide belt. They can sort five to six tons. So they can scale, they can have a volume and they can have a precision output. So it is a problem which help me to understand what the solution should be. And many times it is the customer will actually guide you. Okay, this is what he wants and this is what the market wants. Yeah, unless and until you become like Facebook, where you develop first and then you deploy social things. But again, it had orchid and hotmail earlier. It depends on which domain. If you are into a B two B, I would say business to business. You should always talk to the business. If you are into B two C, it becomes a different part. So in my case, my part was B two B. So I had to go with the business guys. I had to see what are the challenges, what are the workers facing the challenge? So we saw, like to a waste management company, once the truck comes in, the loading, the unloading, where to keep the decision by a supervisor. It was all manual. There was like paperwork, but the paperwork was very shabby, very unorganized. And they were working hard, they were like sweating every day. But they were not able to scale. They didn't have a dignified job or after so much of hard work, after so much of pain, they were not able to scale easy. And that is where we got a solution where they can get a dignified role, dignified job by using this automation solution where they can scale, get volume and their life becomes easy.
PRONOJIT
Got it? Makes sense. Completely. So next I want to hop upon one of my favorite questions, which is how to build competitive modes in deep tech products.
JITESH
Okay? How different are you in the solution and which problem are you solving? Again, it boils down to that thing. See how beautiful your product is or how beautiful the rocket is. Doesn't matter if it doesn't reaches or brings back you the requirement which you started with the vision. So if the problem solves the customers, if he has a technical problem, you are solving it. If the customer needs a scaling problem, if you're solving it, it is in the cost or it is practical. It shouldn't be like if a person needs some Maruti and you're trying to sell him, giving a sales pitch of BMW, though he may like it, but he may not purchase it, though it is a better product. What does the market need? At what price? You can start slow. I always say you can build a smaller product and then you can keep on improving it. Or an MVP, you can start with the first customer. He could be as small or as big, but you go in steps, sell a Maruti, improve it, then make another company, then again improve till you reach to BMW and Audi level. You should have these different features. India is a bigger it is like 1.3 billion. There different customers. You have to start with the smaller stuff and don't feel shy because the smaller part, once you solve it, gives you satisfaction, gives you money, it gives you churning and your product becomes stable. They are your good testers. You don't need a testing team as a startup deploying on the customer and you're working together to solve the problem. If his business grows, it's the best satisfaction you get.
PRONOJIT
Yeah, you may hear some very strict complaints, but at least you get some real hand feedback, right? So how do you see now AIML shaping this waste management industry in the next three to five years?
JITESH
Till now, in US and UK, the sorting waste is different, the people are different. They are developed countries, we are developing nations. The urbanization is growing, so our cities will also grow. When the cities grow, the waste also grows. And today, due to the social ecommerce, many things to reach at the urban level or the villages, the towns which are converting into urban. It means packaging going, waste going, biomedical going, hospitals being developed, every city and the nearby towns also have a similar kind of waste as per the geolocation of India. Now AI comes into picture because the US and UK is sought waste using spectral cameras where they have a clean waste. In India the waste actually has dirt, crossiness and everything and they need a cost effective solution. AI computer vision can help them using normal cameras or some extra sensors which we develop and they get that same volume, let's say three tons per hour, five tons per hour, seven tons per hour. Scanning their waste on a conveyor which runs at 3.2 meters per second by identifying plastic, polymer or glass or metals or paper or cardboard within few millisecond and in that few millisecond also giving the signal to the robotic arm or the pneumatics part to eject it out. So that is where AI comes into picture. And due to the GPU power or due to this, the embedded parts, high end GPUs and high end processing and also AI, the few algorithms they can be run on CPUs also, I think there are very good color sorters available algorithms on OpenCV also. And there are these platforms which are coming up on and NX where you can convert all the models to different platforms so they are helping directly let's say the embedded part also. So they are helping you to actually run the algorithms very faster so that learning curve has reduced. So if you have a hardware, if you are very good understanding the concepts are clear, you can actually keep on trying on cost effective hardware or fast or a smaller hardware and you can build solutions around them. So that is actually changed a lot and we all are aware the hardware and the software will keep on improving. So let's say a few years back the power of an embedded let's say Jetson coming or today or in coming from Nvidia or a GPU let's say 30 80 30 40 Ti coming. It will always have better course, better performance. Or the one CV library or the CC plus everything are improving on both ends. So you will never go down, you will always scale. Your code is always going to work better. Today I seven, I eight, I nine, it won't come down actually got it.
PRONOJIT
So one interesting aspect just come to my thought is right now we have a trend of this large language models, those are open source and there is lots of applications and other downstream applications that are getting built on top. So this is all because we have all the text available freely online and these big cloud providers have the service and all to churn it out coming to you. You have painstakingly built these vision systems by labeling data from initial days and then gathering more and more data which is very proprietary and unique. Do you think maybe that can also be a business owner wherein you give your vision system as a service?
JITESH
Yeah, I think starting with that the last question which you said. Yeah so we are actually signing with some MoUs where the governments across India let's say the South Asian countries what they are trying to do is they had a problem, they want to understand the waste of their towns let's say. So they are asking us to give us the vision in the first phase once they understand the different categories of waste as per the geo location, we build them the plot. Yeah this is happening now, earlier we were giving only the whole product but now everyone is getting smarter. So for the requirement collection also across the towns they are using smaller, you can say, solutions which would be taken by a three car cycle across the South Asian countries and where there would be a vision in a small conveyor where they keep on putting the waste, maybe near the oceans, near the ocean cities or the sea cities. And they want to see what kind of waste is going through. And they will build the plants across. It and this is happening. And there are the second use case of vision where the Mrs in the input and at every step, let's say. There are some 56 screeners or five six filters I would say, just to help. So every filter there is a vision before and after so you will know how is it passing through the hardware. Let's say there's a machine and there's a robot and you haven't been asked the robot to pick. Let's say this pet bottles you can have one more vision in the starting which can guide all the robot saying the robot one you pick the thing. Robot two you pick the thing as per the flow. So this is also happening where the role of the supervisor is being going to be taken by the Aiper. So the decision maker will say okay, you machine, you take this thing. Second machine you do the same.
PRONOJIT
Got it? Sounds interesting. All right, so let's dive now into the last segment of our discussion and you have taken a shitwa from the initial ideation days to a very mature stage with traction right now. So for the listeners of the podcast, what would be your advice on the top three things that they should consider to take their startup or product from zero to one?
JITESH
Okay, the first thing is it should be a genuine and a real problem of a customer. See, either you become a B two B or it's A B two C. Which problem are we solving? It shouldn't be technical first and then the problem. It should be first a problem. And what solution do they want? Let's say it's A, B, two B or a B, two C. Is it scalable? Is there a solution available in the market? If there is a solution, do not reinvent. Please find the gaps of that solution. What are the missing things? It could be a car, let's say. Does a car has those features or not features or any other, let's say a mixer or an AC or a fridge. So what are the gaps in that earlier product? Are the customers happy or not? Once you have that key scalable there are some general problems. Now comes the technology part where you guys are good. See, you will always find a technical answer, trust me. Math and science will always help you to reach and especially once you have clearly written the requirement this is the problems. Once they are fixed by this course I can give them in fact, just starting the course doesn't matter. It's an MVP or a minimal viable product which you can make and showcase to your first customer or first user and they will give you a genuine feedback. If you keep on improving you get a unique product, a product which beats your competitors because you've already done those exercise earlier, not later. So I think these are the simple things. Technology is there, right? I think it's not about technology, it's about problem. We are solving technology. You will find the technology on Google or through some good resources you hire. But if the problem is not genuine, your team will get frustrated. They won't know which way to go.
PRONOJIT
Will not.
JITESH
Once they see the problem, once the team sees a vision, once they see, okay, this is a problem I'm solving. I'm a part of a problem solver. Once you become coming, that more that problem solving. More things becomes easy. Once you have a written problem, this is a simple thing, I would say.
PRONOJIT
Yeah. So I think get the basics done as you showed and invest as much time as possible into that. All right. And finally, you are a technology startup in Gujarat. Now, Gujarat is not generally associated with technology. Later on, deep tech. What is your viewpoint on what's happening and how the technology space in Gujarat is evolving?
JITESH
I think that's a good question. I would wish Gujarat I can see like in the next ten years if manufacturing and software, they both combined, Gujarat would be more powerful. Especially. We heard about the news of semiconductor manufacturing coming into near to Amdapur. And second point is, I do see an evolving part where from services people are more startups are building products related to AI and hardware. I also see more semiconductor manufacturing or semiconductor companies evolving. So there would be a need of software and hardware buys. I see some very good MoUs been done with colleges for the next after COVID so that they can hire for the next five to ten years. I think in the next couple of years we should see some good companies evolving in Gujarat. Very good startups, very good matured companies. And the local Gujarat people, they might not have or maybe they will attract talent over here. See, it's not a Metro over here. It's neither a tire two city, I would say, but it's between tire two and Metro, I would say. And that is the flexibility you get over here. The early days of Pune or Bangalore, you used to get right. So I hope this happens soon. At least in the next five years or seven years down the line. I see some lot of involvement coming.
PRONOJIT
Promising times ahead, I'm sure. All right, Jetesh, thanks a lot for sharing your story and story of ishithwa with us today. I'm sure the listeners have gathered a lot of insights and knowledge which will help them to take more confident steps in their startup journey. Thank you again for taking out time and wish the entire A team of Vishudva a lot of best wishes.
JITESH
Thank you. Thank you for the opportunity. Thank you.