<p>The term ‘AGI’ (Artificial General Intelligence) is commonly used by technology writers, engineers, AI-industry leaders, and philosophers, yet mostly without a precise definition. One prevalent way to characterize such systems is as being capable of performing most tasks at a human level. In this paper, we propose that certain basic cognitive capabilities are essential for AGI systems. These systems would be generally intelligent because they possess a variety of cognitive capabilities, enabling them to reason, be creative, understand the world and perform other high-level cognitive tasks such as learning, planning, complex decision making, and more. Our aim is to explore the most fundamental capabilities necessary for general intelligence and distill the essential building blocks required for AGI systems. Based on this, the main part of this paper is devoted to the claim that analogical reasoning serves as a core capability, underpinning each of these fundamental cognitive capabilities. The paper concludes by presenting guidelines for developing AGI systems that prioritize analogical reasoning as a fundamental architectural principle.</p>