APEX : AI Enhanced Procurement Optimization Agent

• 

  Dashboard

Inspiration

Procurement is critical but often inefficient, relying on spreadsheets and guesswork. We built APEX to transform this with AI, moving from manual processes to intelligent, automated decision-making.

What it does

APEX is a multi-step AI agent that analyzes supplier data, evaluates risks, and generates optimized procurement recommendations. It combines vector search, LLM analysis, and real-time data to automate the entire supplier selection workflow.

How we built it

Backend: Python with TiDB Serverless for data + vector storage AI: FAISS for vector search, GPT-3.5 Turbo for analysis Frontend: Streamlit for interactive dashboard Integrations: CSV export, email/Slack notifications

Challenges we ran into

Transforming structured supplier data into effective vector embeddings Orchestrating seamless data flow between vector search, LLM, and frontend Solving the "cold start" problem with automatic vector index building

Accomplishments that we're proud of

Creating a true multi-step agent that goes beyond simple Q&A Achieving real-world practicality with actionable export features Building a seamless integration between TiDB's vector search and AI analysis

What we learned

Multi-step agents are far more powerful than individual AI components TiDB Serverless is incredibly capable as a unified data+vector platform Actionability is crucial—AI insights must lead to concrete business actions

What's next for APEX : AI Enhanced Procurement Optimization Agent

Integrate real-time market data via Model Context Protocol (MCP) Add predictive analytics for price forecasting Develop multi-agent negotiation capabilities with supplier APIs

Built With

• faiss
• python
• streamlit
• tidb