Views: 44
Imagine a customer service department that never sleeps, never gets frustrated, and can handle thousands of inquiries simultaneously: that’s the power of AI agents in customer service. The agent-AIs are poised to reshape entire digital “industry”, cutting costs and enhancing productivity: hence, customers are witnessing the dawn of a new era in which intelligent machines take center stage in decision-making, automation and innovations.
Background
In 1965, Gordon Moore predicted that the number of transistors on a microchip would double every two years, a prediction that has been since called the Moore’s Law. This principle held for decades, giving rise to fast computing advancements; but a new phenomenon has appeared, promising facilities to accelerate digital innovation at an even greater rate. The rise of AI agents, fueled by ever-smarter algorithms, doubles efficiency not every two years but every six months.
This rapid evolution isn’t just about raw computing power: it is about the rise of AI agents: smart computer programs that can understand what’s happening around them, figure out what to do, and then actually do it to get things done.
Note. The “Moore’s Law”, the principle that transistor counts double roughly every two years, is widely considered to have slowed or ended in its traditional sense due to physical and economic limits. While transistor density still increases, the exponential cost improvements and performance gains have stalled, forcing the industry to pivot towards chiplets, new materials and specialized AI architecture. The “new era of computing” includes: – chiplets and 3D stacking (instead of smaller chips, manufacturers are connecting smaller functional blocks/chiplets together); – specialization: general-purpose CPU gains have slowed, leading to a rise in domain-specific architectures like GPUs and AI accelerators; – alternative materials: research into materials beyond silicon, such as graphene or new lithography technologies, aims to continue performance increases; – software efficiency: as hardware gains slow, optimizing software becomes more critical for performance.
Source: https://www.google.com/search?q=moore%27s+law+is+dead&oq=Moore%E2%80%99s+Law&gs_lcrp
The agent-AIs are poised to reshape entire digital “industry”, cutting costs and enhancing productivity: the humanity is on a brink of a new era of super-intelligent machines capable of automative decision-making.
The agent AI’s deployment is “becoming a common place”: but as many people are being familiar with the “general AI models”, fewer have grasped the power and potentials of a new AI’s “version” – the AI agents, as autonomous digital systems that are far more practical and efficient than familiar GenAI models functioning as “smart assistants”.
An AI agent is not just a piece of software: it is “a self-sustaining system capable of analyzing its environment, making decisions, working with tools and executing actions autonomously”, says Freshworks AI Agent’s website.
Source and citations from: https://www.freshworks.com/freshdesk/ai-agents/scaling-customer-delight/?tactic_id=5627261&utm_source
The AI models have been evolving rapidly during last couple of years: first, there was predictive AI that analyzes data and uses machine learning algorithms to forecast future outcomes. Then, the process moved to generative AI that created new content like text, images and music. Now, the agentic AI stage arrived, which not only generates content, but it is also able to be conversational as well as autonomously act and react.
The “autonomous agents” – apart from the GenAI predecessors – can “reason” based on predictions (made from large datasets), and also “perceive the environment” and take autonomous action; they can even learn from feedback and adapt. Agentic AI (or AI agents) can help executing tasks and are expected to be top strategic technology tools. This digital evolution emphasizes autonomy and adaptability: hence, agentic AI is poised to transform industries like healthcare, finance and manufacturing by seamlessly integrating with data platforms and helping with time-consuming jobs. For example, agent AI can act as a “digital workforce”, making decisions and adapting to new situations with remarkable efficiency.
The new AI agents are built on advanced technologies like machine learning, natural language processing (NLP) and neural networks” they can navigate complex workflows, learn from user interactions and operate in conversational and workflow-driven settings.
Agentic AIs: principles
Proper understanding the difference between agentic AI and GenAI helps clarify that agents are practical implementations of broader agentic AI principles.
However, there are some prerequisites:
a) to define clear objectives: such as needs to be achieved with an intelligent agent. Whether it’s reducing response times, enhancing customer satisfaction, or cutting operational costs, having clear objectives will guide customers/supplies implementation process and help measure success.
b) assessing and preparation of data: AI agents rely on high-quality data to function effectively. Ensure having robust data collection and management systems in place: it includes customer interaction data, transaction histories and other relevant information. Clean and structured data will enable the AI agents to provide accurate and relevant responses.
c) planning using human oversight: while AI agents can handle many tasks autonomously, it’s important to have a plan for human intervention when necessary. Ensure that there are clear guidelines for when and how human agents should step in to assist, providing a safety net for more complex or sensitive interactions.
Some opinions on AI agents
In the last few years, the tech industry has boldly proclaimed that AI “agents” (as the latest buzzword) are going to provide fundamental changes; e.g. in the same way that AI chatbots like OpenAI’s ChatGPT gave people new ways to “surface information”, but AI agents will fundamentally change how people approach work.
But the changes depend on how one defines “agents,” which is not an easy task. Much like other AI-related jargon (e.g. “multi-modal,” “AGI,” and “AI” itself), the terms “agent” and “agentic” are becoming diluted to the point of meaninglessness. That threatens to leave OpenAI, Microsoft, Salesforce, Amazon, Google, and the countless other companies building entire product lineups around agents in the digital realm. For example, an AI agent from Amazon isn’t the same as an AI agent from Google or any other vendor, and that’s leading to confusion and customer’s frustration.
Reference to: Zeff, M. & Kyle W. “No one knows what the hell an AI agent is”. 2025. – TechCrunch. Archived from the original on March 18, 2025.
Other source: Purdy M. “What Is Agentic AI, and How Will It Change Work?”. 2024. – Harvard Business Review. ISSN 0017-8012.
The AI agent’s definition is varied: e.g. in March 2025, OpenAI published a blog post that defined AI agents as “automated systems that can independently accomplish tasks on behalf of users.” Yet, the company released developer’s documentation that defined agents as “LLMs equipped with instructions and tools.”
Leher Pathak, OpenAI’s API product marketing lead, later said in a post on X that she understood the terms “assistants” and “agents” to be interchangeable; the note that just further complicated the content. Meanwhile, Microsoft’s blogs tried to distinguish between agents and AI assistants: the former, which Microsoft calls the “new apps” for an “AI-powered world,” can be tailored to have a particular expertise, while “assistants” merely help with general tasks, like drafting emails.
Anthropic says that agents “can be defined in several ways,” including both “fully autonomous systems that operate independently over extended periods” and “prescriptive implementations that follow predefined workflows.” Reference to Zeff, M. and Kyle W. (2025).
Another company -Salesforce – has had perhaps the most wide-ranging definition of AI “agent”: according to the software giant, agents are “systems that can understand and respond to customer inquiries without human intervention”. The company’s website lists six different categories, ranging from “simple reflex agents” to “utility-based agents.”
Note. Salesforce is the world’s leading cloud-based Customer Relationship Management (CRM) platform that helps businesses manage, track and analyze “in complex” customer data, sales processes and marketing efforts; it acts as an integrated “operating system” for companies, utilizing AIs to connect sales, service, marketing and corporate teams.
Reference to: https://www.salesforce.com/eu/?ir=1
More companies are reaping the benefits of generative AI agents, which can draw from trusted customer data and quickly providing the valuable insights. Done manually, that kind of data gathering and analysis can cost time and money: but AI agents can do that instantly, freeing up employees to work on more complex and critical issues.
AI agents can do so much more than just data analysis and instant customer care: with AI agents in place, companies are able to scale teams quickly, hit key performance indicators, and solve problems before they become major issues. And the new technologies are opening other exciting possibilities.
There are two presently types of promising Agent-AIs:
= Utility-based agents. These agents use a utility function to make decisions. They can evaluate different actions based on an expected utility measure to choose the optimal approach. This model is ideal when there are multiple solutions to a problem, and the agent needs to decide on the best one, such as an autonomous car deciding on the safest and quickest route.
= Goal-based agents. These powerful tools are tailored to achieve specific goals. They consider the consequences of their actions and can make decisions based on whether they can use the action to achieve its objective. This means they can navigate incredibly complex scenarios autonomously and respond to the environment through sensors.
Source and citations from: https://web.archive.org/web/20250318205426/https://www.salesforce.com/agentforce/what-are-ai-agents/?bc=DB
Suggestions for the AI Agent structure
Huang K. proposed an AI Agent reference architecture, which consists of seven interconnected layers, where each layer is building on the functionality of other “layers”:
Layer 1: Foundation models – provide the core AI engines to power agent capabilities.
Layer 2: Data operations – manage the complex data infrastructure required for AI agent operations, including Vector database, data loaders, RAG.
Layer 3: Agent frameworks – sophisticated software and tools that simplify the development and management of the AI agents.
Layer 4: Deployment and infrastructure – provide the robust technical foundation for running AI agents.
Layer 5: Evaluation and observability – focus on assessing the safety and performance of AI agents.
Layer 6: Security and compliance – a crucial protective framework ensuring AI agents operate safely, securely, and conform regulatory boundaries. At this layer security and compliance features embedded into all the AI agent stack layers are integrated together.
Layer 7: Agent ecosystem – represents the AI agents’ interface with real-world applications and users.
Source and citation: Huang K. Agentic AI: theories and practices. 2025. Cham: Springer. ISBN 978-3-031-90025-9.
Enhancing customer’s operational efficiency: examples
= In healthcare: AI agents are improving operational efficiency and enhancing patient care by leaps and bounds. Workflow agents are being used to analyze medical images, flag potential issues for review by human doctors, and even predict patient outcomes based on vast datasets of historical medical records. Conversational agents, on the other hand, are transforming patient care. They can act as virtual health assistants, reminding patients to take their medication, answering questions about symptoms, and even conducting initial triage to determine if a patient needs to see a doctor urgently. The combination of these agents is particularly powerful in remote patient monitoring. Workflow agents can analyze data from wearable devices, detecting anomalies and predicting potential health issues. If a problem is detected, a conversational agent can reach out to the patient, gather more information and -if necessary- alert healthcare providers or emergency services.
= Intelligent process automation in finance: the financial sector, with its complex workflows and stringent regulatory requirements, is ripe for disruption by AI agents. Workflow agents are taking center stage by automating everything from data entry and reconciliation to risk assessment and fraud detection. For example, in the mortgage approval process, workflow agents can gather and verify applicant information, assess credit risk and even make preliminary approval decisions. They can flag unusual cases for human review, ensuring that complex or high-risk applications would receive the necessary scrutiny.
Meanwhile, conversational agents are revolutionizing personal banking: they can handle balance inquiries, facilitate transfers and even provide financial advice based on a customer’s spending habits and financial goals. The integration of these two types of agents creates a seamless, efficient banking experience being available around the clock.
The AI agent’s future
The “digital world” is presently at the brink of a new AI’s revolution: and the future belongs to those who can harness the power of AI agents. From customer service to healthcare, finance to retail, etc. these digital entities are reshaping industries and redefining possible solutions.
The new AI agent’s evolution is the beginning of a larger path: as the AIs are continuing their exponential growth, the digital community is able to witness the emergence of even more sophisticated AI-agents which can not only “perceive and act”, but also “learn and adapt” in real time. These AI agents can even collaborate with humans in increasingly complex ways.
The developers that can successfully deploy the new types of AI agents will not only increase corporate efficiency and reduce operating costs; they would fundamentally transform the operating framework in business and create “entirely new paradigms of work” and digital-human’s interaction. Thus, the so-called agent-powered AI future is opening huge possibilities and almost limitless spheres of corporate opportunities and customers’ satisfactions…
Agentic AI is well-positioned to revolutionize the corporate world and governance though the compatibility with already existing executive systems, the ability to create personalized user experiences, as well as in robust security features which are so indispensable presently.
Such AI agent-innovations will basically transform numerous modern industries such as sales, services, marketing and commerce.
Gartner predictions (which trace how AIs are influencing business systems and decisions) revealed recently that “by 2028, about 15 percent of day-to-day work decisions would be made autonomously through agentic AI, up from a zero point in 2024”. More than that, the AI “would infiltrate the B2B procurement: by 2028, almost all B2B-buying would be AI agent intermediated, pushing over $15 trillion of B2B spend through AI agent exchanges”. Procurement would be re-programmed, not by policy but by invisible agents: i.e. traditional search engine optimization (SEO) and pay-per-click (PPC) “will give way to agent engine optimization”, products will need to be machine-readable and procurement will shift to efficient, autonomous machine-to-machine transactions.
Source and citations from: https://www.gartner.com/en/articles/strategic-predictions-for-2026
A new corporate future is approaching: with the AI agents that can seamlessly handle complex customer inquiries, adapt marketing strategies in real-time, and optimize supply chains with unparalleled efficiency. The potential for agentic AI to streamline operations and enhance customer experiences is immense. One of the most exciting aspects of agentic AI is its ability to learn and improve over time. As these AI agents accumulate more data and experience, their decision-making abilities will become increasingly sophisticated. This continuous learning process will help businesses stay ahead of the curve, responding quickly to market changes and customer needs. The integration of agentic AI with data platforms will be a game-changer, providing seamless access to vast amounts of information and enabling more informed and timely decisions.
Reference and citation from: https://web.archive.org/web/20250316071541/https://www.salesforce.com/agentforce/what-is-agentic-ai/?bc=WA#whats-next-for-agentic-ai
More on scaling AIs in: https://www.gartner.com/en/articles/scaling-ai
Besides, some AI agents’ developers are already introducing legal rules that shift final responsibility onto platform’s users: it means that AI agents do not grant any “legal eligibility” in using services. This approach is now “cautioning users” against relying on AI-generated content for decision-making!
Practical information
Vertex AI Agent Builder is an open and comprehensive platform that empowers enterprises to rapidly build, scale and govern corporate entities by “implanting” AI agents in the company’s data. It provides the full-stack foundation and extensive developer needed choices to transform applications and workflows into powerful and reliable agentic systems at global scale.
Reference to: https://cloud.google.com/products/agent-builder?utm_source=google&utm_medium=cpc&utm_campaign=Cloud-SS-DR-GCP-1713666-GCP-DR-EMEA-EMEA-en-Google-SKWS-MIX-na&utm_content