Opinion
Agentic AI in organizations: How to prepare for the next generation
AI agents work independently to achieve pre-defined goals, upgrading automations to a whole new level. This enables a wide range of applications, including task automation, autonomous customer support, IT management, employee experience management, and customized marketing and sales.
Agentic AI, the next generation of artificial intelligence centred around autonomous agents executing tasks independently, is gaining traction in organizations.
According to Gartner, a third (33%) of organizational applications will contain agentic AI elements by 2028.
Research company IDC has also recently published the results of a survey performed among technological leaders in various organizations which found that 84% of them see the organizational use of AI, mostly relying on agentic AI, as the central technological investment within the organization.
AI agents work independently to achieve pre-defined goals, upgrading automations to a whole new level. This enables a wide range of applications, including task automation, autonomous customer support, IT management, employee experience management, and customized marketing and sales.
These systems comprehend broad contexts, make decisions, perform tasks, learn and improve from past operations, and adapt to changing circumstances. They take the initiative, aiming to solve complex challenges and achieve the desired results while showing high-quality autonomous behavior when independently tackling issues, or “true intelligence” when performing situational analysis, leading to better decision making. Agentic AI marks a leap in quality when compared to traditional systems, which work according to rigid rules, lacking in adaptability. It allows for smart end-to-end automation of complex processes - from supply chain management to customized customer care. Imagine a system that autonomously identifies business opportunities, independently communicates with clients, and makes necessary adjustments on its own - all without the need for a human in the loop.
In addition to well-established players such as ServiceNow and Salesforce, more specialized companies like Alsera, Beam AI, and Ampcome are also active in this space.
The difference between AI Agents and Agentic AI
While the term “AI agents” is commonly used in the industry, it's important to distinguish between a single agent operating as a component within a system, and true Agentic AI—a holistic system that coordinates multiple, function-specific agents to complete complex tasks. Such agents do more than mere “task execution” and can be categorized into several primary types: General information agents which use RAG databases in non-regulatory environments; Hardcoded information agents that operate based on predefined rules; Execution agents which perform actions across various systems; Assistant agents which chat with clients and Management agents which plan and manage the actions of other agents. These agents rely on classical machine-learning techniques as well as on LLMs to identify anomalies, predict scenarios, and maintain optimal action.
What does the use of Agentic-AI look like?
Similar to using an AI assistant, it usually begins as either a text or voice chat. After receiving the user’s request, the system turns to automatically plan and execute the process. The system then uses the management agent to allocate goal-oriented tasks to the different agents. It is often the case that after receiving the agents’ replies, the system prompts the user for additional information or clarification. When the preparation process completes, the system executes the plan, monitors activity, automatically readjusts, learns, and completes the conversation with the user.
Embedding Agentic AI in Organizations
To identify areas that would most benefit from an Agentic-AI system, it is best to focus on high-volume processes - areas requiring a large number of tasks to be performed regularly and repeatedly over time. A high degree of repeatability, meaning tasks that repeat in a very similar manner with minimal exceptions requiring human reasoning, and have a direct impact on the customer experience or the organization’s operational efficiency, are also key factors. Increased speed, accuracy, or availability can create immediate value, whether through cost reduction or increased user satisfaction. The chances of achieving a rapid increase in business value rise when all three of these conditions are met.
Common areas that benefit significantly from such improvements include customer service, technical support, client and employee onboarding, stock management, and internal organizational coordination. According to Alsera, organizations that have adopted Agentic-AI platforms managed to automatically resolve up to 75% of their service calls, reduce support costs by up to 90%, and increase customer satisfaction by up to 85%.
A key consideration is whether to develop such a solution internally or utilize an existing platform. While developing simple RAG systems (which provide answers by retrieving relevant information) can sometimes suffice, the development of Agentic AI systems is increasingly complex and costly. Today, commercially available systems offer access to sophisticated algorithms, robust process management, information security, and ongoing maintenance. These platforms provide a fast, secure, efficient, and flexible solution for most organizations, enabling them to focus on value creation rather than development.
Yoel Jacobsen is the CTO of EMET Group.
