A central challenge in digital marketing has been to draw a direct, quantifiable line between content creation on social media and e-commerce sales. This paper introduces the AETE, a multi-agent system designed to autonomously generate and promote content on external social media platforms to drive sales growth for products on Amazon. The AETE operates through a closed-loop, sequential architecture of four specialized AI agents. The Market & Audience Insight Agent (MAIA) begins by analyzing a target product’s Amazon page, particularly user reviews, alongside external social media trends to identify key selling points and define the target audience profile. Based on these findings, the Content Strategy & Creative Agent (CSCA), which serves as the system’s reinforcement learning core, formulates a strategic “content angle” and outputs a detailed “creative brief.” Next, the Multi-Platform Content Generation Agent (MCGA) executes this brief, producing adapted, multi-modal content for platforms like TikTok, YouTube, and Instagram, complete with embedded affiliate tracking links. Finally, the Performance & Attribution Analysis Agent (PAAA) analyzes traffic and sales data via third-party APIs, such as affiliate programs, to calculate the return on investment (ROI) for each content strategy. This ROI is fed back to the CSCA as a direct reward signal, enabling it to optimize its strategy through experience. The core contributions of this research are twofold: 1) A fully automated workflow that transforms raw user comments into actionable creative briefs. 2) The formalization of content marketing strategy as a reinforcement learning problem, where the “content angle” serves as the fundamental unit of action, allowing for quantifiable learning and iteration. This study presents a new paradigm for building performance-oriented, autonomous marketing systems that can operate effectively in public, uncontrolled digital environments.