Giuliano Neroni - IoT, AI, and Blockchain: The Infrastructure of the Future

Okay, thank you very much for being here today everyone. We are at the stage Capella Napole 2025 has been just started. We will have three days of event full of energy, full of networking, full of Napolitan food. I hope you enjoy it. Uh we will start right now with Juliano Juliano Neoni.

Um she he is the founder of Growa Blockchain and AI developer. uh winner of the contest such as ECP accelerator in Lugano. Uh his talk is how IoT, Internet of Things, AI proc uh preprocessing and blockchain oracles can work together to create transparent datadriven systems. So let's make a round of applause to Juliano and wish you the best.

Hi guys. Nice to meet you. I'm Juliano Neoni, CTO and founder of Global Services Consultancy. Uh we are a Kadari company under the minister of communication information technology in Kadar. I'm happy to be in Napoli at NATH to talk about the synchronized trinity how the blockchain AI could work together to be applied in a real world application.

First of all, I want to start from a problem. Uh never like like now we have uh so much technology applied and the majority of the time this technology have such a problem. The 73% of integrated tech solution fail because the majority of the time they are not build to together. Uh the problem is that we keep build separate tools and platform that that communicates. So blockchain operates in a silo AI in another one and the IoT in a third one.

So that make uh uh that the client the one that have to use the platform will uh instead of find a solution will find only problems. So imagine three different assistant that have to work in uh in some on something that then at the end of the day doesn't work because uh someone do something the other one do another thing and at the end they will not work together. So the clients end up with a lot of apps, a lot of interface, a lot of data. We have data everywhere but at the end we find something that doesn't work. But when blockchain, when AI and when IoT we start working together in this uh uh this time the result will be exponential because the difference that they could act is that they could learn could adapt and could be optimized continuously.

We have the IoT devices that for us are the hands are the ears are the uh eyes because it's the one that let us to uh gathering data let us to distate the sensor everywhere. It is the one that the most important thing that permit to process automation. But how we could process the automation? This is something that we do with the with the eye. The eye is our brain.

The eye is the one that allow us to do data analysis. Is the one that allow us to understand pattern. Is the one that continues to learn. But the most important thing is the blockchain. Especially when we you build something that that have to work with institutional partner because the the blockchain is the one that allow you to have a data integrity.

That is the one that allow you to have a trustless automation is the one that allow you to follow the compliance for someone is more important for someone less is also the one that give you the decentralized security. IoT devices are the smart object that react instantly when something happen. Is the one that understand how for example in our case in our case how the greenhouse the state of the green houses the state of the of the weather and all kind of things and then based on that is something that could act but AI is the cognitive center. uh we have to think something. Um there is a a a huge difference between our brain and the AI system.

The both of them work exactly on the experience because the eye you will give them the data set and they will start to learn based on that the our brain work in the same way but the difference is in the subjective. The our brain is subjective. A I not AI when founding something when recognize a pattern we always follow the same solution and is the one that understand when something happen if this is wrong uh based on the supervision training we start to say how to act differently. Uh one of the biggest problem is that the majority of the the majority of the time the data are not clean. they are the raw data are very noisy and fragmented.

This is where they help us to understand how manage this kind of data. Uh for example in our system in our architecture we have four different kind of layer. The first layer is the one that allow us to do data validation and anomaly is the one that allow us to clean the data and to bring the data to the second layer. The second layer is the when we are going to find the pattern. A pattern is when with our neural network we understand if there is something that is similar to a situation that I know and after this tape was done there is the third layer is the one that apply the output of the data to the predictive modeling and it's the one that let us uh understand what is the decision to take and optimize the decision.

But the most important layer is the first one. It's the one that bent of the confidence scoring decide to act or not because is the one that let us to understand what is happening around and so understand if this decision had to take or not. AI doesn't just process data it creates value with with these four things. predicting problems, optimizing resource allocation, learning from failures, and discovering in the patterns. But at the end of the day, the blockchain is the most important thing because it's the one that that let us work with the data that are trusted, verified, and recorded.

And this is very important, especially if you work with a solutional partner. But what is the problem of the blockchain? The majority of the time the blockchain are too slow and very expensive for to be applied to a real world application. Uh you have to think that when uh you have an application for an industrial partner you have thousand of me session per hour and this is something very expensive and also the speed is very important. When someone say that number are not important trust me guys number are very important.

Uh let me say that a traditional blockchain usually for a transaction need three three three minutes and also the gas cost also if you think that is very low based on a huge amount of uh transaction we are talking about a huge amount of money uh for example we found our partner in ICP I don't know how much of you know the internet computer product blockchain but is the one that let us uh work let us find the solution solution that could be applied in a real world application. Uh I don't know much you know about SAP but they have a a kind of application that give us a high speeded computerational platform. I like to say that is like the an evolution from the app to the feedback optic. I don't know how many of you have the age that know how the idea lab work but this is exactly the same kind of evolution. The workflow is very easy IoT sensor detect change AI process is instantly at the blockchain executives and this is something that happen in uh in second not in hours.

Uh let me do an example P on our real case. Uh and uh our our uh company work in the agriculture sector in the ASG sector and also in automation based on data in this sector. uh for example if we have a detection from IoT devices in a greenhouse that understand that the level of volutric content that let the conductivity at this time of the day with this kind of aable transpiration going very low the eye start to decide the data will go directly to the AI uh there will be an inference on that bas of this kind of inference where this data will pass to a a material operation This material operation understand what is happening and talk directly with the blockchain and give the information to the blockchain and then the block the blockchain save the information give the information back to the I that decide to to accept the the operation if there is any an operation to ex otherwise will not do anything and only record the data this is something that could be you will think that take a lot of time but is something that will take around 45 Second, uh we have to think to the this kind of ICP that I advise any developer to try to use and to apply to this kind of application because it's something that could change your way to apply the web three uh world to a real work case and uh this is something that let us to build a smart software agents. Uh what I want to say with that is something that give us a layer capability memory memory persistence inter canister communication a real time processing each canister remember every action whatever whatever happen and then also thanks to the outall that could do outside is something that let us to predict what could happen next and it's something that all will be recorded it's something that happen. It is something that let us to create a neural network that not not only uh will will be managed in a in a private structure but is something that uh let us to create an oracle based on that.

We have three kind of oracle based on that an IoT data oracle that is the one that permit to work with the data validates and uh validate the time stamp the sensor data. There is the AI processing oracle that is the one that exages the machine learning algorithm is the one that between the call in input and output let us to bring data and take data from a side and there is the satellite oracle. This one is the one that integrate external data but let after they they will be green they will be a trustable data and this is something that all running natively on on blockchain. So what is the problem that is solve on chain oracle eliminates trust gap and single points of failure. Um but now I want to talk with a with the I don't I don't know many developer will be there when we when we build something we have to start to build something with a a user oriented vision.

We are used to build something not with the with thinking to the needed of the clients but we build something the majority of the time think with our brain. This is the common mistake because at the end of the day who need to use that architecture is the client and so the uh the majority of the time we are building something that for them is very difficult to use something that will be very difficult to use and is but the the user need to have something that have a very huge complexity underline but the but for them will be very easy to use. I want to this is this is an architect that could be used everywhere in a sector where you need a realtime decision could be applied to manufacturing to optimization energy and management and um I want to close with a with a with a small phrase our challenge is to make the complex simple and impossible possible. Thank you guys and uh if you have any question I will be there to reply.