Agentic Supply-Chain Optimization with Mastra AI and OpenAI o3 Reasoning
Supply chain operations often fail due to decoupled policies between logistics, warehousing, and procurement. Inspired by the 'SCOPE' research, this challenge requires you to build a multi-agent coordination system using Mastra AI. You will leverage the high-reasoning capabilities of OpenAI o3 to manage 'coupled policies'—where the decision of one agent (e.g., procurement) directly affects the constraints of another (e.g., shipping). You will implement a Mastra Workflow that orchestrates these interactions, utilizing persistent memory to track inventory states and dynamic tool calling to interface with simulated ERP systems.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a multi-agent supply-chain coordinator using Mastra AI and OpenAI o3 to solve coupled policy optimization problems.
How work is evaluated
The system must resolve a simulated supply-chain bottleneck where shipping capacity is reduced by 40% while demand remains high.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
Constraint Adherence
Ensures no daily shipping exceeds the max_shipping_per_day limit.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Optimization Efficiency
How close the agent's cost is to the theoretical minimum. • target: 95 • range: 0-100
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Orchestrate complex agentic workflows using Mastra AI's graph-based state management
Implement coupled policy logic where agent outputs serve as dynamic constraints for subsequent reasoning steps
Apply OpenAI o3 reasoning models to resolve conflicting multi-objective optimization problems in logistics
Design and integrate custom tools for Mastra agents to interact with transactional data
Master the use of Mastra's built-in memory for long-running stateful agent sessions
Reference links and supporting material
Synthetic supply chain data including real-time inventory levels, vendor lead times, and fluctuating shipping costs simulated in a JSON format.
Challenge setup
The system must resolve a simulated supply-chain bottleneck where shipping capacity is reduced by 40% while demand remains high.
Challenge input
JSON object with inventory, current orders, and shipping capacity.
Mastra AI
Used for building the multi-agent graph and memory-safe workflows.
OpenAI o3
Provides the high-level logic needed for coupled policy optimization.
Evaluated output
Consolidated Logistics Plan (JSON).
- Ensures no daily shipping exceeds the max_shipping_per_day limit.
- How close the agent's cost is to the theoretical minimum. • target: 95 • range: 0-100
- Benchmark: Supply Chain Resilience Benchmark
- Optimization Efficiency target: 95
- 1 public reference case
- Docker execution harness
- Python sandbox (unavailable on Versalist)
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
- OpenAI · Optional
- Mastra AI · Optional
- o3 · Required
- Mastra AI · Optional
Evaluation contract
- Constraint Adherence · Weight 1
- Optimization Efficiency · Weight 1
Metrics
- Optimization Efficiency · Target 95
Benchmarks
- Supply Chain Resilience Benchmark
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON object with inventory, current orders, and shipping capacity.
Output: Consolidated Logistics Plan (JSON).
Execution harness: Docker
Evaluation tools
- Mastra AI · Orchestration Framework
- OpenAI o3 · Reasoning Model
Recipe freeze marker
No recipe freeze time is recorded.