Hello, everyone. My name is Shubham Gupta, and I am an analytics and AI leader at Noblesoft Solutions. First of all, I would like to thank Conf42 for inviting me as a speaker for Cloud Native 2025 conference. I'm very excited to talk about my topic, which is agentic AI and blockchain autonomous systems for next generation innovations. So folks, let's dive in. I will begin my presentation by, referring to some key points from Sam Altman's, three observations. Sam Altman is the CEO of OpenAI, the company responsible for creating ChatGPT. So he says that AI intelligence scales with resources. So the more computing power, the more data and the more inference you provide to AI systems, the smarter and more intelligent they become. Another key point. is that the cost of AI drops by a factor of 10 every year, which means that AI is becoming cheaper and also getting commoditized and is no longer accessible only to big tech companies as it was happening a few years ago. So what this means for 2025 and beyond, right? So AI is no longer playing a passive role in the decision making process, but it is slowly becoming a virtual co worker who can actively collaborate with you. And we are already seeing this disruption in software engineering already. Now, what is agentic AI? agentic AI refers to autonomous software agents capable of decision making with minimal human intervention. These agents do not merely react, but they proactively execute strategies to solve complex tasks. what is agentic AI again, right? agentic AI is the future of autonomous intelligence. Now, as you can see in the diagram, it has six key components, context awareness, goal setting, processing, decision making, ethics and governance, and then autonomous action. Autonomous action is a very important word here, right? these AI agents operate on the core principle of autonomous operation, goal driven behavior, context awareness, and self improvement. They feature an autonomy engine that enables self initiated actions and resource management. these AI agents use adaptive learning. to continuously evolve through reinforcement learning and pattern recognition. It incorporates a decision matrix for real time risk assessment and scenario simulation. It maintains ethical boundaries through a dedicated governance layer. It integrates with other AI agents, through a collaborative ecosystem. It also represents the next evolution in AI, which is moving from reactive to proactive intelligence. here is the day to day use case of, AI agents. AI agents can proactively take actions here. They can post, the content on different social media platforms. They can decide what to post, when to post, where to post, how to post, so that you can get maximum engagement on your content. So you can see that, these AI agents are proactively making decisions. so now let's learn a little bit about AI agents, right? So they have six core components. The number one is learning system. so these AI agents build comprehensive knowledge base and then it adapts its learning method continuously. Then comes goal framework. So the framework sets very clear objectives. It takes progress against these. targets or objectives, and it ensures that alignment happens with those intended goals. Then comes reasoning engine. reasoning engine processes complex logical operations. Um, it makes decisions based on available data. It solves problems through systematic analysis. It maintains consistency in logical flows. Then comes memory system. So the memory system manages short term information storage. It maintains long term knowledge retention. It handles efficient information retrieval. It integrates various memory types seamlessly. Then comes interaction layer. this layer manages communication interfaces. It processes API requests. It generates appropriate responses. and then finally, we have safety protocol, this protocol implements ethical frameworks. It enforces, constraint systems, validate actions and outputs, and it also ensures safe operation boundaries. Now, let's dive into the agentic AI architecture. in this architecture, we have six key steps again. Number one is learning. Then comes adoption. Then comes action. Then comes anticipation. Then we have efficiency. And then refinement. learning, like we discussed, it constantly ingests new data. and it's training itself continuously. dynamically updating its knowledge base. When it comes to adoption, the architecture implements continuous feedback loops. When it comes to actions, the system demonstrates initiative and task execution, generating strategic responses to the challenges, and capitalizing on the opportunities when detected, right? Then comes anticipation. So through forecasting capabilities, risk assessment protocols, and trend analysis, The system stays ahead of emerging patterns and potential challenges. Then comes efficiency. So the architecture maintains peak performances through optimization of processes, smart resource allocation, and continuous performance tuning. And then finally we have refinement. So the system ensures excellence, operational excellence through continuous improvement cycles, dynamic adjustment capabilities, and rigorous quality assurances. Now, let's see how these AI agents work, right? So it starts with input processing. So the system processes natural language inputs from users. It also interfaces with the external system. through API request, then we have knowledge based integration, where these agents step into comprehensive domain knowledge. it understands user specific context for personalization. Then comes task planning. So the agent analyzes specific goals to achieve. It breaks down complex tasks into sequential steps. Is it then efficiently allocates available resources and then it says task priorities based on importance and urgency, right? then comes reasoning engine, right? So the system applies logical inferences to draw conclusions. It identifies patterns and complex data sets. Then comes tool integration. So the agent connects with external tools through APIs. It processes data using specialized processors. It leverages external services for enhanced capabilities. It executes custom functions for a specific task. Then comes, execution engine, right? So the, System org, is designed to handle multiple tasks simultaneously. It handles errors with fallback options. It maintains state across operations. It validates results for accuracy. Then comes response generation. the agents select optimal, response formats. It ensures quality. It determines the best delivery methods. And then it incorporates user feedback for improvements. Then comes system monitoring. So the agent takes key performance metrics. It monitors for potential errors. It analyzes users usage patterns. It maintains system health checks. And then finally, we have security layer. so here, system verifies user authentication. It manages authorization levels. It ensures data privacy compliance. And it also maintains detailed audit logs. Now, here is another way to explain how AI agent works, right? we have input sources. So the system ingests data through three primary channels. number one is knowledge based integration for foundational information. Number two is user queries for direct interaction. And number three is API calls for external system communication. So then each input stream feeds into the central processing pipeline, which is AI processing pipeline. So now In this stage or in this layer, the agent processes incoming information through sophisticated analysis systems. It parses natural language, structures data formats, and prepares input for deeper processing. Then comes, reasoning layer. So the cognitive core applies advanced reasoning capabilities. to the analyzed data. It synthesizes information from multiple sources, identifies patterns, formulates logical pathways to solution, and this forms the basis for decision making. And then finally, you have, decision making, which is, The agent evaluates possible actions based on process information. It weighs alternatives, consider constraints, and selects optimal approaches. This ensures that the response aligns with the user needs and system capabilities. Then comes, task execution. Here, the system implements chosen actions through its It's execution engine, it manages resource allocation, handles concurrent operations, and maintains operational consistency throughout the process. Then comes agent collaboration. So multiple AI agent systems work in concert to achieve, complex goals. They share information, coordinate actions. and leverage collective capabilities for enhanced problem solving. Then come response generation. So the final output synthesizes results from all previous stages. It then formats responses appropriately, ensures clarity and accuracy, and delivers information to the optimal channel for user confirmation. So now let's talk about the difference between AI automation, AI workflows, and AI agents, right? So AI automation, as we have been seeing in the past, it just executes predefined scripts with some fixed rules, right? So Zapier email integration could be an example for this. Then comes AI workflows. These workflows manages complex processes with defined decision points. So a good example here could be Netflix recommendation system. And then finally comes AI agents. these agents create autonomous solutions through intelligent reasoning. One example of this could be automated inventory management and supply chain management. And we will go over this, example a little bit later. So now, after going over, AI agents, now let's dive into the blockchain fundamentals. So a blockchain is a distributed e routable ledger that records transactions across multiple nodes in a peer to peer network. Now, what is the benefit of using a blockchain, right? So the main benefits are decentralization, meaning the information is not centrally stored. It is scattered across several different nodes. Data, once recorded, cannot be changed. the system is transparent, meaning everyone, accessing the data. sees the same information and it's safe and secure because of, consensus mechanisms, right? again, why blockchain, right? it removes single point of failure, it ensures data integrity and transparency, it enables trustless environments, and then it facilitates, secure peer to peer transactions. And the key strengths of blockchain, namely decentralization and immutable record keeping are very important when data must be verifiable and tamper proof across multiple stakeholders, right? so here is a blockchain architecture, right? here we can see that there are different types of blockchains. We have consortium blockchains, we have public blockchains, and then we have private blockchains. These blockchains use different consensus mechanisms. when they need to add a new block in the blockchain, right? So it can either be proof of work or it could be proof of a stake. Now, we have gone through AI agents. We have, understanding about blockchain. So now let's see why we should combine AI and blockchain together, right? So AI and blockchain can work together, very well to achieve autonomous innovation, which is where the future is heading, right? first of all, data integration for AI. So AI models often rely on very large data sets. So blockchain here ensures data accuracy, data quality, authenticity, which actually, make these AI models more reliable. Then comes autonomous agent coordination. So for example, for complex tasks, we might have many decentralized agents. that needs to coordinate with each other, without a central authority. So in this case, combining AI and blockchain can work very well. then come smart contracts for decision enforcement. So agentic AI decisions can be executed automatically via smart contracts. Thereby reducing the chance of fraud, right? And then come auditability of AI decisions. So immutable records of decisions made by AI agents can be stored on the blockchain for future auditing. so all these, characteristics make this combination a really good combination for autonomous innovation. So here we can see that, AI. is provided with extensively real world data, which it uses to generate insights, and then it triggers on chain transactions, through smart contracts. Then these smart contracts, record the transactions on the blockchain, making them available to all stakeholders, right? here we can see that once the transactions are recorded onto the blockchain, all stakeholders access the same information no matter where they are situated. so now let's see some important use cases of, AI and blockchain innovation together, right? So it can very well be used in supply chain management where AI agents optimize logistics and blockchain provides, immutable record keeping. Then we can use AI agents and blockchain combo and decentralized finance. We can use them in healthcare data sharing and then energy grid optimization. Now, as I mentioned before, let's go over the AI and blockchain and supply chain management. So enterprise systems and IoT sensors generate data, which is then fed into the AI engine. or AI agents to generate insights and alerts. Now, based on these insights, a smart contract is triggered to record payment on the blockchain, ensuring all stakeholders have access to the same real time data. now in this case, you can see we have manufacturers, suppliers, retailers, and then we have transport providers, then we have warehouse. So all of them are accessing the same copy of data, which is coming from blockchain ledger. and it is the source of truth for everyone here. Now, when it comes to feature outlook, we can use, AI and blockchain in AI use case, where AI competition happens very close to the. Source of data, it can help in federated learning where AI runs on multiple devices without exposing sensitive data. Now, AI innovations, what is the road ahead for us, right? So by 2035. one person may have access to the intellectual capacity of an entire 2025. This is crazy, right? and then new form of work will emerge. very different from today's job. So it will transition from knowledge based economy to an allocation based economy. AI will be integrated into everything and then smart will become the new default. So with these pointers in mind, I would like to conclude my session. I would like to thank all of you for listening to my presentation and I wish you a good day. Thank you.