What is Agentic AI? Key Features, How It Works, Benefits, Limitations, and Use Cases

Agentic AI refers to artificial intelligence that can act independently, making decisions and executing tasks without constant human input. As this technology advances, it holds the potential to transform industries by reducing costs, improving decision-making, and increasing scalability.

Unlike traditional AI systems that rely heavily on human input, agentic AI possesses the ability to understand context, make independent decisions, and learn from experiences.

In this blog, you'll gain a clear understanding of what is agentic AI, how it works, and its practical applications.

Ready to explore the potential of Agentic AI? Dive into its applications and learn how to implement it in your projects with upGrad’s Artificial Intelligence and Machine Learning courses. Gain hands-on experience through 15+ case studies and projects. Enroll today and take your AI expertise to the next level!

What is Agentic AI? Comparison with AI Agents

Agentic AI enables systems to take independent, decision-critical actions, learn from experiences, and adapt based on real-time data. This capability allows it to operate effectively in high-stakes environments, where quick decisions are crucial, such as in medical imaging, autonomous vehicles, and financial trading systems.

For example, in autonomous vehicles, it continuously analyzes traffic patterns and adjusts the vehicle's speed, route, and braking decisions in real-time, ensuring safety and efficiency.

The true strength of Agentic AI lies in its capability to:

• Automate complex tasks: In healthcare, Agentic AI systems have demonstrated high diagnostic accuracy. For instance, AI models have achieved up to 96.3% sensitivity and 93.3% specificity in analyzing digital pathology images.
• Scale operations: AI's ability to process vast amounts of data enables organizations to scale operations efficiently.
• Reduce human error: By operating without human biases or fatigue, itI provides consistent and accurate results. For example, AI-driven coding systems have reduced coding errors by up to 35%, showcasing their potential to transform the healthcare industry by improving billing accuracy and operational efficiency.

With the increasing reliance on Agentic AI for complex decision-making and automation, professionals with expertise in this technology will be highly sought after. To build your expertise in this field, check out these top-rated courses:

• Masters of Science in Artificial Intelligence and Machine Learning
• Executive Diploma in Machine Learning and AI with IIIT–B
• DBA in Emerging Technologies with Concentration in Generative AI

To better understand Agentic AI, let's explore its evolution and key milestones in recent advancements.

History and Recent Developments in Agentic AI

Agentic AI's advancement traces back to early AI developments that focused on task automation. The first wave of AI concentrated on narrow applications, such as rule-based systems that followed a defined set of rules. As computing power and machine learning algorithms improved, it began to develop more autonomy. 

Key milestones include:

• Early AI systems in the 1950s and 60s focused on logic and problem-solving but required human intervention for guidance.
• The rise of machine learning in the 1980s and 90s allowed AI systems to "learn" from data and improve autonomously over time.
• The introduction of reinforcement learning (RL) has been crucial in the development of Agentic AI, as it allows systems to learn from the consequences of their actions, making them more effective at decision-making without human intervention.
• Recent advancements: In the 2020s, Agentic AI has found applications in autonomous vehicles, real-time stock trading, and AI-powered healthcare diagnostics. Innovations like self-driving cars, intelligent assistants, and autonomous drone deliveries have highlighted the enormous potential in practical scenarios.

Recent developments also focus on explainable AI (XAI), which helps ensure that decisions made by Agentic AI can be understood and interpreted by humans, even when the systems are operating autonomously.

Also Read: AI - A Comprehensive Timeline of Artificial Intelligence

Having looked at Agentic AI’s journey, let's examine its differences from Generative AI to better understand their unique roles and functions.

Difference Between Agentic AI and Generative AI

While both Agentic AI and generative AI use machine learning and deep learning techniques, they have distinct functionalities and applications. 

To provide a clearer perspective, let's explore a detailed comparison of the key differences and features:

Also Read: AI vs. Human Intelligence: Key Differences & Job Impact in 2025

While Agentic AI and Generative AI differ in their core functionalities, it's essential to also distinguish between Agentic AI and AI agents, which serve distinct roles.

Difference Between Agentic AI and AI Agents

AI agents refer to software systems that act on behalf of users, performing tasks that are repetitive or require decision-making with a degree of human oversight. If you're wondering what is AI agent, it typically refers to systems that assist with specific tasks but still rely on human direction. In contrast, Agentic AI can operate fully autonomously, making decisions and taking actions without human intervention.

Here's a breakdown of how Agentic AI and AI agents differ: 

Also Read: Rational Agents in AI: A Detailed Guide on How They Function

Now that we understand the differences, let’s explore the mechanics of how Agentic AI works in a structured, step-by-step approach.

How Does Agentic AI Work? A Step-by-Step Guide

Agentic AI works by gathering real-time data from sensors or IoT devices, processing it using advanced algorithms like machine learning and deep learning, and making autonomous decisions. Each step is crucial for ensuring optimal performance in complex environments. 

Below is a step-by-step breakdown of how this process unfolds.

1. Data Collection (Perception)

The first step in the workflow is gathering data from the environment through sensors or input sources like IoT devices. With projections of over 75 billion IoT devices globally by 2025, the scale of real-time data is growing rapidly. This surge in data plays a crucial role in improving decision-making accuracy. 

For example, in autonomous vehicles, sensors and cameras capture data to detect obstacles and traffic signals, forming the foundation for AI’s decision-making. The integration of 5G technology allows this data to be processed and transmitted instantly, enhancing AI's effectiveness in dynamic environments.

Examples:

• Autonomous Vehicles:
  Issue: Navigating through complex traffic is challenging due to varying conditions and obstacles.
  AI Intervention: Perception systems use radar, LIDAR, and cameras to analyze the surroundings in real time.
  Outcome: Ensures safe navigation, avoiding accidents, and making split-second decisions in dynamic traffic environments.
• AI in Healthcare:
  Issue: Continuous monitoring of health metrics is difficult without immediate access to accurate data.
  AI Intervention: Wearables like smartwatches collect and analyze health data in real time.
  Outcome: Enables quicker diagnosis, early detection of health issues, and more efficient patient care.

Also Read: What Is Data Collection? Explore Types, Methods, Steps & Challenges

2. Data Processing (Reasoning)

After data is collected, Agentic AI processes it using advanced algorithms like machine learning and deep learning. This enables the system to identify patterns, predict events, and refine its decision-making. Techniques such as deep neural networks and transformers like BERT and GPT help AI systems understand context.

These techniques are especially useful in industries like retail and finance. They improve decision-making in areas such as customer service. Edge computing plays a key role by reducing latency. This enables real-time data processing.

Examples:

• AI in Retail
  Issue: Retailers struggle with inventory management and predicting customer buying trends.
  AI Intervention: AI analyzes customer behavior data and forecasts buying trends in real-time.
  Outcome: Inventory levels are automatically adjusted, reducing stockouts and overstocking, and improving sales.
• AI in Healthcare
  Issue: Healthcare providers face challenges in identifying early signs of health risks, such as heart attacks.
  AI Intervention: AI processes patient data from wearable devices and electronic health records to detect early warning signs.
  Outcome: Timely predictions of potential health risks lead to earlier interventions, improving patient outcomes and reducing emergency incidents.

Also Read: Top 5 Data Analysis Courses to Pursue Online in 2025

3. Decision-Making

After processing the data, Agentic AI evaluates potential actions based on insights derived from its reasoning phase. Using deep Q-learning, it refine their strategies by learning from previous actions, continuously improving their performance in dynamic environments such as financial markets or emergency traffic management.

The ability of this modern AI to predict, adapt, and act on real-time information without delay transforms how industries handle high-stakes decisions.

Examples:

• Autonomous Vehicles:
  Issue: Traffic conditions change rapidly, requiring quick adjustments in speed and lane positioning.
  AI Intervention: The AI system continuously monitors the road and adjusts speed or changes lanes based on real-time traffic data.
  Outcome: The vehicle maintains optimal speed and safety, responding effectively to dynamic traffic conditions.
• AI in Finance:
  Issue: Financial markets are volatile, requiring quick decisions for profitable trades.
  AI Intervention: Trading algorithms analyze market conditions and autonomously execute trades, adjusting strategies based on fluctuations.
  Outcome: The AI system maximizes profits by capitalizing on market trends and minimizing losses.

Also Read: Decision Making Tools and Techniques: A Quick Guide

4. Action

Following decision-making, Agentic AI implements actions through software, robotics, or autonomous vehicles. Advances in robotics and IoT have expanded AI’s ability to act in the physical world. For example, in smart cities, AI integrates with infrastructure to optimize traffic management and energy consumption. 

AI can adjust traffic lights or re-route vehicles to reduce congestion, enabling quicker and more efficient responses to urban challenges.

Examples:

• Robotic Process Automation (RPA):
  Issue: Manufacturing processes are often slow and prone to human error.
  AI Intervention: AI-driven robotic systems autonomously manage tasks like assembly line optimization and quality control.
  Outcome: Significant reduction in human involvement, faster production times, and higher accuracy in manufacturing processes.
• Smart Cities:
  Issue: Urban areas struggle with traffic congestion, inefficient public transport, and energy waste.
  AI Intervention: AI autonomously controls traffic flow, adjusts street lighting, and coordinates public transportation systems.
  Outcome: Improved traffic efficiency, reduced energy consumption, and smoother operation of urban infrastructure, enhancing overall city management.

Also Read: 30 Exciting RPA Projects Ideas for Beginners in 2025

5. Feedback Loop (Learning)

The feedback loop is where Agentic AI refines its actions based on past outcomes. By evaluating its decisions and the results, the system adapts its strategies for continuous improvement. This learning process is essential for enhancing performance over time. 

Techniques like federated learning enable decentralized data processing, allowing AI systems to learn from diverse datasets across hospitals, factories, and financial institutions, while maintaining privacy and security.

Examples:

• AI in Healthcare:
  Issue: Diagnosing complex medical conditions often involves a high degree of variability and human error.
  AI Intervention: AI systems analyze past diagnostic outcomes and medical records using machine learning techniques.
  Outcome: The system improves predictive accuracy, leading to quicker and more accurate diagnoses.
• AI in Customer Service:
  Issue: Customers often experience inconsistent and generic responses during support interactions.
  AI Intervention: AI agents, which are designed to autonomously handle customer inquiries, use feedback from previous interactions to refine their responses and adapt to individual needs. If you're wondering what is AI agent, it's an intelligent system that can learn, adapt, and interact with users autonomously.
  Outcome: The support becomes increasingly personalized, improving customer satisfaction and loyalty.

Also Read: How Does Generative AI Works and it Application

Having understood how Agentic AI works, let’s look at the advantages it brings and the potential challenges you should consider.

Benefits and Limitations of Agentic AI

Agentic AI brings powerful benefits like real-time decision-making, pattern recognition, and the ability to automate complex tasks. It enhances efficiency in high-stakes environments, such as healthcare and autonomous vehicles. However, limitations include dependence on vast data and potential biases in algorithms. For example, in autonomous vehicles, AI systems can misinterpret environmental data, leading to safety concerns. 

Below are key benefits and limitations of Agentic AI.

Also Read: 5 Significant Benefits of Artificial Intelligence [Deep Analysis]

Having discussed the benefits and challenges, let’s explore how Agentic AI is being used in diverse sectors.

Practical Use Cases of Agentic AI Across Various Sectors

Agentic AI is rapidly advancing, making a significant impact across industries from healthcare to retail. By automating complex tasks, improving decision-making, and adapting to dynamic conditions, it is changing how businesses operate. These applications go beyond traditional AI uses, offering innovative solutions to challenges in various fields. 

Here are some key use cases across different sectors:

1. Healthcare

In healthcare, Agentic AI is not just improving operational efficiency but also enhancing patient outcomes through real-time data processing and autonomous decision-making. AI systems are now capable of analyzing complex medical data and providing actionable insights, often without requiring direct human intervention. This is reshaping medical practices, allowing for more personalized and timely care.

Emerging Use Cases in Healthcare:

• Automated Medical Imaging Analysis: Agentic AI platforms, like Zebra Medical Vision, autonomously analyze medical images, such as X-rays and MRIs, to detect anomalies and diseases with remarkable accuracy, enabling early diagnosis and reducing human error.
• AI in Surgical Planning: AI agents assist in the planning of surgeries by reviewing patient data and recommending optimal procedures. For AI agents examples, PathAI helps pathologists analyze pathology slides for more accurate cancer diagnoses, improving patient care outcomes in real-time.
• AI-driven Patient Monitoring Systems: AI-powered systems predict health deteriorations in real-time by continuously monitoring patient vitals. These systems help healthcare providers detect potential issues early, enabling timely interventions and enhancing preventive care.

• What is AI Agent in Healthcare?

  In healthcare, what is AI agent often refers to assistive tools that support doctors. These include systems that suggest treatment plans or monitor patient vitals.

  AI agents examples include PathAI for pathology diagnostics and Babylon Health for AI-powered consultations, both aiding clinicians by providing instant insights from patient data.

2. Finance

The finance industry is increasingly adopting Agentic AI to streamline operations, manage risks, and offer tailored services. By processing vast amounts of market data in real-time, It helps financial institutions predict market trends, assess risks, and enhance customer experiences.

Emerging Use Cases in Finance:

• AI in Risk Management: Institutions like Moody’s use AI to analyze large datasets, predicting financial risks and improving credit ratings accuracy. This helps mitigate potential losses by considering a wide array of market and economic factors.
• Customer Behavior Prediction: AI agents, such as those used by Ant Financial, analyze transaction history and online activity to predict consumer behavior. This allows for personalized financial advice, boosting engagement and customer satisfaction.
• Fraud Detection: AI models detect fraud by analyzing transaction patterns and identifying anomalies that may indicate suspicious behavior. Real-time analysis of spending habits helps flag fraudulent activities before transactions are completed.
• Personalized Investment Advice: AI tailors investment recommendations by analyzing user profiles, including risk tolerance and financial goals. By learning from user interactions, these systems provide dynamic portfolio management, adapting to changing market conditions.
• Quick Insight: AI Agents in Fintech: Financial tools powered by AI often use agents to perform tasks like fraud detection, customer segmentation, and portfolio suggestions. AI agents examples in finance include robo-advisors like Betterment and chatbots like Erica by Bank of America. These are practical answers to what is AI agent in the finance industry: focused, task-driven assistants that improve service efficiency.

3. Supply Chain and Logistics

Agentic AI is revolutionizing supply chain management and logistics by automating processes, improving route optimization, and managing inventory with high precision. These autonomous systems ensure that goods are delivered efficiently and on time, reducing operational costs and enhancing customer satisfaction.

Emerging Use Cases in Supply Chain and Logistics:

• Autonomous Warehousing: Amazon uses AI-driven robots for sorting, picking, and packing. Agentic AI also optimizes supply chain routes in real-time, considering weather, traffic, and inventory levels, reducing delivery times and costs.
• Smart Route Optimization: AI agents are utilized to optimize delivery routes for companies like Amazon, ensuring faster and more cost-effective deliveries. AI-powered autonomous trucks and drones adjust routes in real-time to save fuel and time.

4. Retail and E-commerce

In the retail and e-commerce sectors, Agentic AI is being used to enhance customer experience, streamline operations, and personalize marketing strategies. By leveraging AI to analyze consumer data and automate tasks, businesses are able to deliver personalized experiences that increase customer engagement and satisfaction.

Emerging Use Cases in Retail and E-commerce:

• Personalized Shopping Experiences: AI systems like Shopify’s analyze consumer behavior to offer personalized product recommendations, enhancing the shopping experience. This drives higher conversion rates, boosts customer satisfaction, and increases repeat purchases.
• Autonomous Inventory Management: Walmart uses AI to track inventory, predict demand, and restock products autonomously, ensuring shelves are always stocked with the right items. This reduces stockouts, optimizes inventory costs, and provides a smooth customer experience, ultimately boosting sales and efficiency.

Also Read: Top 13+ Artificial Intelligence Applications in 2025

5. Future Outlook of Agentic AI

Looking ahead, Agentic AI is poised for even broader adoption and deeper integration across industries. As AI technology advances, its ability to perform complex tasks autonomously and learn from experience will only improve, opening up new possibilities in every sector.

Trends Shaping the Future of Agentic AI:

• Increasing Autonomy: Agentic AI will take on more sophisticated tasks in sectors like healthcare, finance, and logistics. For example, AI-driven robots in logistics, such as those used by FedEx, handle inventory and route optimization autonomously, increasing efficiency and reducing human oversight.
• Enhanced Learning and Adaptation: AI systems will become more adaptable, continuously learning from new data. In healthcare, IBM’s Watson for Oncology is already personalizing cancer treatment plans and will soon adjust them in real-time based on patient progress.
• Wider Industry Adoption: Agentic AI will expand into fields like agriculture, education, and government. In agriculture, John Deere’s AI systems optimize crop management by predicting yields and adjusting irrigation, while in education, AI platforms like upGrad offer personalized learning experiences.
• Ethical Considerations: As AI takes on more autonomous roles, ensuring fairness and transparency in its decision-making will be critical. For example, AI used in finance for credit scoring, like Zest AI, must avoid bias and ensure decisions are justifiable to customers and regulators.

Also Read: Future Scope of Artificial Intelligence in Various Industries

Having explored its use cases across sectors, let’s explore the crucial steps and best practices to get started with Agentic AI.

How to Get Started with Agentic AI: Key Steps and Best Practices

To implement Agentic AI successfully, begin by defining clear objectives that align with your business goals. Identify the tasks AI will handle, such as automating processes or improving decision-making. Ensure your infrastructure supports real-time data processing. Selecting the right data sources and algorithms is essential, as they directly influence the AI’s performance and ability to make accurate, adaptive decisions. 

Below are the essential steps to guide your implementation.

1. Understand Your Business Needs

The first step in implementing Agentic AI is understanding your business objectives and identifying specific tasks that require autonomous decision-making. Unlike traditional AI, It excels in managing complex, real-time environments where rapid, adaptive decision-making is crucial. 

For instance, automating workflows, optimizing supply chains, or personalizing customer interactions. Clearly defining these tasks will help align your AI strategy with business goals, ensuring efficiency and effectiveness in deployment.

Key Considerations:

• Define objectives: What do you aim to achieve with Agentic AI like improve customer experience, automate internal processes, or enhance predictive analytics?
• Evaluate scalability: Ensure the chosen AI solution can grow with your business as demands evolve.

Example: If you aim to streamline customer support, deploying an AI-powered chatbot like Zendesk Answer Bot can automate responses, reduce wait times, and improve user satisfaction.

2. Select the Right Tools

With the variety of Agentic AI solutions available, it’s crucial to choose those that align with your specific requirements. Focus on tools that offer Agentic-specific capabilities like real-time feedback loops, which allow AI systems to adapt and refine their actions over time, and explainability, which helps ensure transparency in decision-making. 

Consider factors such as ease of integration, scalability, and the core functionalities you need, like machine learning, real-time data processing, or natural language understanding.

Key Considerations:

• Platform compatibility: Choose platforms that integrate seamlessly with your existing systems for a smoother deployment.
• Tool functionality: Ensure the tools support key features, such as real-time adaptive learning, decision-making autonomy, predictive analytics, or explainability, depending on your use case.

Example: For AI in healthcare, tools like PathAI for image recognition or TensorFlow for building custom machine learning models can be great choices depending on your application.

3. Prepare Data

Agentic AI’s effectiveness heavily relies on the quality of the data it processes. Unlike traditional AI, It requires real-time, labeled, and contextual datasets to make autonomous decisions in dynamic environments. 

This phase involves cleaning inaccuracies, organizing data, and ensuring it’s suitable for training and continuous analysis. Data needs to be structured in a way that allows AI to understand context, adapt to changing conditions, and improve over time.

Key Considerations:

• Data quality: Ensure that data is accurate, consistent, and well-organized, as errors can negatively impact decision-making.
• Real-time and labeled data: Use datasets with real-time updates and clear labels to allow AI systems to make quick, informed decisions.
• Data diversity and context: Include diverse datasets that reflect real-world scenarios and provide context, helping the AI make more relevant and effective decisions.

Example: In fraud detection, financial AI systems rely on real-time transaction data that is labeled to spot patterns of fraud. These datasets help the system recognize subtle changes in behavior and detect fraudulent activities as they occur, ensuring timely intervention.

4. Pilot Testing

After preparing your data and selecting tools, it's time for pilot testing. This phase allows you to test how well the AI performs in real-world scenarios, enabling you to address potential issues before full deployment. Conducting pilot tests with real-world data ensures that the AI solution meets your business goals and functions as expected.

Key Considerations:

• Test with real-world data: Use data that mirrors the challenges the AI will face during actual operations, ensuring it can handle real-world complexities.
• Monitor performance: Continuously track AI performance using relevant metrics to ensure it aligns with your goals. Important metrics include:
  • Reaction latency: The time it takes for the AI to process data and take action.
  • Accuracy drift: Monitor how the AI’s accuracy may change over time, ensuring it doesn’t degrade.
  • Adaptability: Evaluate the AI’s ability to learn and adjust based on new data or unexpected changes in its environment.

Example: Autonomous vehicle companies like Waymo test AI systems with real-world driving data, monitoring metrics such as reaction latency (how quickly the system responds to road conditions) and accuracy drift (how well the AI adapts to new or evolving traffic scenarios) to ensure safety and efficiency before public rollout.

5. Continuous Monitoring and Optimization

Continuous monitoring and optimization are crucial for long-term success in this field. As the system interacts with new data, ongoing assessment ensures it stays effective and adapts to changing environments. Key areas to focus on include monitoring autonomy thresholds, where AI decisions extend beyond predefined parameters. 

Additionally, addressing learning decay is essential, as system performance can decrease over time. Regular updates and adjustments are essential for maintaining efficiency, accuracy, and alignment with business goals.

Key Considerations:

• Ongoing evaluation: Regularly assess AI performance, tracking when it exceeds or falls short of autonomy thresholds.
• Learning decay: Watch for performance degradation and retrain the system with fresh data when necessary.
• Refinement: Continuously refine AI models using new data to enhance accuracy and effectiveness.

Example: In e-commerce, Amazon’s recommendation engine optimizes product suggestions based on customer behavior. By monitoring AI adaptability and making necessary adjustments, it maintains personalized experiences while avoiding performance drift.

Also Read: 34 Artificial Intelligence Project Ideas With Source Code

6. Best Practices for Implementing Agentic AI

To ensure the effective implementation and ethical use of Agentic AI, follow these best practices:

• Data Privacy and Ethics: Ensure that data collection and processing comply with privacy regulations. AI systems should operate transparently and be free from biases, particularly in sensitive areas like healthcare and finance.
  Example: IBM Watson Health ensures that its AI healthcare applications comply with ethical standards and patient privacy regulations.
• Interdisciplinary Teams: Collaboration across diverse teams, including AI experts, data scientists, and industry professionals, ensures that the system aligns with technical standards and regulatory requirements.
  Example: Google DeepMind’s healthcare division works closely with doctors, ethicists, and engineers to responsibly implement AI in medical practices.
• Model Governance and Fallback Protocols:
  Implement strong model governance to monitor AI performance and decision-making, along with fallback protocols for critical scenarios where AI may fail or make errors. This ensures human oversight and enables corrective action when needed.
  Example: In autonomous vehicles, companies like Waymo have fallback protocols that enable human intervention if the AI system fails, ensuring safety in unpredictable situations.

Also Read: AI Ethics: Ensuring Responsible Innovation for a Better Tomorrow

Now that you’re familiar with the basics, take your understanding further with expert-led courses to truly master Agentic AI.

Become an Expert in Agentic AI with upGrad!

Agentic AI is transforming industries by enabling systems to make decisions and perform tasks autonomously, without human intervention. However, many professionals face challenges in mastering this complex technology, from understanding its core principles to implementing it in practical scenarios. 

With practical insights and hands-on projects, upGrad’s courses equip you to build AI solutions for industries like healthcare, finance, and autonomous vehicles. 

Here are some free courses to help you start your journey in the AI sector.

• Fundamentals of Deep Learning and Neural Networks
• Introduction to Generative AI
• Learn Python Libraries: NumPy, Matplotlib & Pandas
• ChatGPT for Developers
• Analyzing Patterns in Data and Storytelling

Struggling to utilize the full potential of Agentic AI? upGrad offers personalized career counseling to help you choose the best path as per your goals. You can also visit your nearest upGrad center to gain hands-on experience through expert-led courses and real-world projects.

Expand your expertise with the best resources available. Browse the programs below to find your ideal fit in Best Machine Learning and AI Courses Online.

Best Machine Learning and AI Courses Online

Discover in-demand Machine Learning skills to expand your expertise. Explore the programs below to find the perfect fit for your goals.

In-demand Machine Learning Skills

Discover popular AI and ML blogs and free courses to deepen your expertise. Explore the programs below to find your perfect fit.

Popular AI and ML Blogs & Free Courses

Reference Links:
https://www.ft.com/content/1b336414-3319-4c2b-adb6-d90426b877c3        
https://arxiv.org/abs/2306.07999    
https://medtechintelligence.com/feature_article/ai-and-human-oversight-a-new-era-in-reducing-medical-billing-errors/    
https://www.researchgate.net/figure/nternet-of-Things-IoT-connected-devices-from-2015-to-2025-in-billions_fig1_325645304    
https://www.nccoe.nist.gov/iot