Autonomous Process Operations Team with CrewAI
Process automation facilities suffer from critical skill shortages and complex shift transitions across operations, maintenance, and safety departments. You will create a multi-agent process operations squad using CrewAI where specialized agents (OperatorAgent, MaintenanceAgent, SafetyOfficer) collaborate to resolve chemical process drift. The squad must reach consensus on corrective actions within 3 interaction turns.
What you are building
The core problem, expected build, and operating context for this challenge.
Orchestrate a role-playing team of collaborative AI agents using CrewAI to automate chemical refinery shift handovers and process drift management.
How work is evaluated
Evaluates multi-agent consensus output quality and task turn count.
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.
unanimous_approval_check
Verifies that Operator, Maintenance, and Safety agents all explicitly approved the final plan.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
agent_interaction_turns
Number of round-trip agent messages to reach consensus • target: 2 • range: 1-6
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Design multi-agent role configurations in CrewAI with specific goals, backstories, and tools
Implement sequential and hierarchical task delegation among process operations agents
Integrate simulated chemical reactor telemetry analysis tools into CrewAI task executions
Evaluate multi-agent consensus reliability and resolution efficiency
Reference links and supporting material
Continuous batch chemical reactor logs with simulated exothermic drift incidents.
Challenge setup
Evaluates multi-agent consensus output quality and task turn count.
Challenge input
JSON containing reactor_id, pressure_psi, temp_c, and catalyst_flow_rate
CrewAI
Required framework assignment.
Evaluated output
JSON with consensus_action, operator_approval, maintenance_approval, safety_approval, and total_turns
- Verifies that Operator, Maintenance, and Safety agents all explicitly approved the final plan.
- Number of round-trip agent messages to reach consensus • target: 2 • range: 1-6
- Benchmark: Multi-Agent Industrial Consensus Benchmark
- Agent Interaction Turns target: 2
- 1 public reference case
- Python 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
- CrewAI · Required
- crewAI · Optional
Evaluation contract
- unanimous_approval_check · Weight 1
- agent_interaction_turns · Weight 1
Metrics
- agent_interaction_turns · Target 2
Benchmarks
- Multi-Agent Industrial Consensus Benchmark
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON containing reactor_id, pressure_psi, temp_c, and catalyst_flow_rate
Output: JSON with consensus_action, operator_approval, maintenance_approval, safety_approval, and total_turns
Execution harness: Python
Evaluation tools
- CrewAI · Multi-Agent Framework
Recipe freeze marker
No recipe freeze time is recorded.