Generative AI for SmartShoe
Drexel University, LeBow College of Business (coursework)
Context
A generative multi-agent prototype for a simulated retail supply chain, built for the Generative Agentic AI Simulation course.
Business question
Can one conversational interface replace manual coordination across demand forecasting, supplier management, inventory allocation and production flow?
Data
Excel knowledge bases for demand forecasting, supplier management, inventory allocation and production flow.
Approach
- A routing agent dispatches each inquiry to specialist agents grounded in their own Excel knowledge bases.
- An orchestrator synthesizes a single recommendation, and significant findings escalate to stakeholders by email.
- Assessed reliability through structured prompts, agent accuracy evaluation and decision-path outcome analysis.
Key insight
Recommendation
Route each inquiry to specialist agents, synthesize one recommendation through an orchestrator, escalate significant findings to stakeholders by email, and test reliability with structured prompts and decision-path analysis.
Impact
-
Prototype (opens in a new tab)
multi-agent generative AI prototype on StackAI with GPT-5 agents
Deliverable
Tools
- StackAI
- GPT-5
- Excel