Multi AI Agents System for Digital Marketing

Multi-agent systems feature throughout Dr. Alfred Ang's working week, both in the software Dr. Ang builds and in the way that software gets written. A full-stack AI developer, Agentic AI solution architect and corporate trainer, Dr. Ang is CTO of Tertiary Infotech Academy Pte Ltd in Singapore, the training organisation behind Tertiary Courses, and builds intelligent agents and business platforms end-to-end across front end, back end, databases, cloud deployment and native mobile apps. More than 300 public repositories on GitHub show the range. Dr. Ang has published Claude Code skills, modular and transferable capabilities that an agent can load on demand, covering tasks such as starting any app, generating README files, pushing to Docker Hub and drafting LinkedIn posts, plus skill sets that produce training slides, learner guides and lesson plans in a consistent house style. Several Streamlit applications built on the Claude Agent SDK generate complete course documents from structured inputs. An AI-powered content management system on Next.js uses the same SDK for content generation, a weekly auto-blog scheduler and a self-improving chatbot. A video-production pipeline routes a request through separate research, scripting, rendering, verification and publishing stages, and an ERP prototype passes work between AI agents and human approvers. On n8n, Dr. Ang has also built retrieval-augmented chatbots, a stock-analysis Telegram agent and a guarded HR assistant. Much of this code was produced through structured vibe coding, with agents, sub-agents and context files doing the heavy lifting under close human review.

This course teaches that workflow directly. It opens with modern agent foundations: what a modern AI agent is, agent skills as modular capabilities, agent memory systems, the fundamentals of the Model Context Protocol, and the step from a single capable agent to a team of collaborating ones. The second topic explains vibe coding and a structured vibe coding workflow, context engineering for reliable behaviour, the main vibe coding tools, and how to add agent skills, the same techniques Dr. Ang uses to keep large projects consistent. Learners then build a collaborative multi-agent system with the OpenAI Agents SDK, using structured outputs, API tool calling and supervisor routing with sub-agent delegation, and deploy it to Streamlit. The exercise is then repeated with the Gemini Agent SDK, again deployed to Streamlit, so that learners can compare the two approaches honestly. The final topic goes deeper into MCP and tool orchestration and into designing hierarchical and sub-agent systems that stay understandable as they grow. Dr. Ang teaches through live builds and frequent checkpoints, showing how a clear context file or a well-scoped skill prevents agents from drifting, and when it is wiser to keep a human in the loop. Questions about cost, reliability and testing are welcomed throughout. Participants leave with deployed multi-agent applications built on two platforms and a repeatable method for growing them safely. To browse the skills, agents and Streamlit apps referenced here, visit Dr. Ang's portfolio at https://github.com/alfredang