Perishable Supply Chain Optimization with GPT-5 Pro & CrewAI
Multi-agent system to track grocery pallets with sensors, this challenge focuses on building an agentic system for predictive supply chain optimization for perishable goods. You will create a multi-agent team that can analyze simulated IoT sensor data (temperature, humidity, location) from pallets, predict potential spoilage or delays, and recommend proactive interventions. Your solution will use CrewAI to orchestrate a team of specialized agents, powered by GPT-5 Pro for general reasoning and Claude Sonnet 4.5 for focused data interpretation. The agents will integrate with a simulated ERP system to retrieve product details and log actions, enabling intelligent decision-making to minimize waste and improve delivery efficiency.
What you are building
The core problem, expected build, and operating context for this challenge.
Orchestrate a role-based agent team with CrewAI, using GPT-5 Pro and Claude Sonnet 4.5, for predictive perishable supply chain optimization.
How work is evaluated
The evaluation will assess the agent team's ability to accurately detect issues in simulated pallet data, predict spoilage risk, and recommend appropriate, timely interventions, while effectively utilizing different LLMs and interacting with a simulated ERP system.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Integrate simulated IoT sensor data streams for real-time analysis.
Design a multi-agent system using CrewAI for supply chain monitoring and optimization.
Connect agents to a simulated ERP system for data retrieval and action logging.
Develop agents capable of predicting spoilage and recommending interventions.
Utilize GPT-5 Pro for high-level decision-making and Claude Sonnet 4 for detailed data interpretation.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The evaluation will assess the agent team's ability to accurately detect issues in simulated pallet data, predict spoilage risk, and recommend appropriate, timely interventions, while effectively utilizing different L...
Challenge input
JSON object with 'pallet_id', 'product_type', 'sensor_data' (list of {'timestamp', 'temperature', 'humidity', 'location'}), 'product_metadata' (e.g...
Agent execution
No agent tool is listed for this challenge.
Evaluated output
JSON object with 'pallet_id', 'spoiled_risk_level', 'predicted_spoiled_by', 'recommended_action', 'justification', 'erp_log_entries'
- The system correctly identifies high/medium/low spoilage risk.
- The recommended action is appropriate for the given risk.
- The simulated ERP system is correctly accessed/updated.
- Prediction Accuracy target: 85
- Python execution harness
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
No tool records are attached.
Evaluation contract
- The system correctly identifies high/medium/low spoilage risk.
- The recommended action is appropriate for the given risk.
- The simulated ERP system is correctly accessed/updated.
- Agents demonstrate coordinated workflow as defined by CrewAI.
Metrics
- Prediction Accuracy · Target 85
- Recommendation Timeliness · Target 15
- Resource Efficiency · Target 300
Workflow tasks
- Task 1
Input: JSON object with 'pallet_id', 'product_type', 'sensor_data' (list of {'timestamp', 'temperature', 'humidity', 'location'}), 'product_metadata' (e.g., 'shelf_life_days', 'optimal_temp_range')
Output: JSON object with 'pallet_id', 'spoiled_risk_level', 'predicted_spoiled_by', 'recommended_action', 'justification', 'erp_log_entries'
Execution harness: Python
- Task 2
Input: Same as PredictiveSpoilageDetection.
Output: Log of agent communications and tool calls.
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.