The goal of this project is to build prototype AI agents that can automate specific workflows using OpenAI's agent framework. These agents will perform tasks such as customer support, internal process automation, or knowledge retrieval using LLMs (Large Language Models). Students will a develop tools, instructions, and orchestrations that power intelligent, autonomous decision-making within business operations.

Each student is expected to complete 60 hours of practical work focused on the following:

• Conducting a use case analysis to determine which workflows are most suitable for agent-based automation.
• Developing and documenting agent components, including the model logic, tools (functions/API hooks), and guardrails (e.g., input validation or security filters).
• Designing a single-agent or multi-agent system, depending on complexity, using code-first orchestration principles.
• Building prompt templates and clear instructions that guide agent behavior effectively.
• Conducting basic tests and evaluations of agent behavior, accuracy, and fallback strategies.
• Delivering a demo or walkthrough of the agent prototype and submitting a final implementation package with documentation.

Project Outcomes and Deliverables

• A working prototype AI agent or agent system, built using Python (and optionally OpenAI's Agents SDK).
• A written implementation guide detailing model selection, tools used, instructions configured, and orchestrations defined.
• A presentation or screen-recorded demo showing the agent in action and describing its workflow.
• A brief report assessing the prototype’s performance and areas for improvement or scaling.
• Optional: A comparison analysis showing how different orchestration models (single-agent vs. multi-agent) affect the outcome.