Skip to content
All work

AI & Data

AI Purchasing Agent

A rational purchasing agent, audited by a team of LLM agents.

The problem

Buy inventory well when prices are volatile, and explain whether the strategy is working.

How I built it

  1. 01Simulated a rational agent using exponential smoothing and a threshold-based purchasing policy, tracked with descriptive statistics.
  2. 02Layered a CrewAI + Gemini multi-agent system that audits the agent's decisions and writes strategic recommendations.
  3. 03Grounded the auditors with a RAG tool that pulls live market data from an external site.

Outcome

  • Decisions are reviewed against live market context, not just historical data.

Next project

IT Audit & Anomaly Detection Suite