Automate Enterprise Workflows with MCP Agents
Building concept of workflow automation agent, this challenge focuses on building a sophisticated multi-step workflow automation system. You will design and implement an agentic solution that can analyze existing enterprise processes, identify bottlenecks, and propose optimized, AI-driven solutions. The system must integrate with various simulated enterprise APIs using an MCP-enabled tool integration layer, allowing it to gather data, execute actions, and report on performance improvements. Emphasize the agent's ability to engage in extended thinking and utilize adaptive reasoning budgets to handle complex, non-linear workflows.
What you are building
The core problem, expected build, and operating context for this challenge.
Build an intelligent agent using Semantic Kernel with Gemini 3 Pro for enterprise workflow analysis and optimization, leveraging MCP tool integration and adaptive thinking.
How work is evaluated
The evaluation will assess the agent's ability to correctly parse a workflow description, identify inefficiencies, propose a coherent optimized workflow, and demonstrate successful (mock) tool integration.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Analyze and model complex enterprise workflows using AI.
Propose optimized, AI-driven workflow improvements with performance metrics.
Utilize Semantic Kernel's planning capabilities for multi-step automation.
Design and implement MCP-enabled tool integration for mock enterprise systems.
Reference links and supporting material
Sample data for 3 tasks
Challenge setup
The evaluation will assess the agent's ability to correctly parse a workflow description, identify inefficiencies, propose a coherent optimized workflow, and demonstrate successful (mock) tool integration.
Challenge input
{"workflow_description": "Textual description of a workflow, e.g., 'Customer onboarding: Sales creates lead, marketing sends welcome email, custome...
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{"bottlenecks": [{"step": "string", "reason": "string"}, ...]}
- Agent identifies at least two distinct and plausible bottlenecks in the provided workflow.
- Agent suggests actionable, relevant steps for optimization.
- The agent successfully calls and receives a response from a mock enterprise API via MCP.
- Workflow Optimization Score target: 80
- 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
- Agent identifies at least two distinct and plausible bottlenecks in the provided workflow.
- Agent suggests actionable, relevant steps for optimization.
- The agent successfully calls and receives a response from a mock enterprise API via MCP.
Metrics
- Workflow Optimization Score · Target 80
- Tool Integration Accuracy · Target 90
Workflow tasks
- Task 1
Input: {"workflow_description": "Textual description of a workflow, e.g., 'Customer onboarding: Sales creates lead, marketing sends welcome email, customer support configures account, billing sends invoice.'"}
Output: {"bottlenecks": [{"step": "string", "reason": "string"}, ...]}
Execution harness: Python
- Task 2
Input: {"current_workflow": "string", "identified_bottlenecks": [{"step": "string", "reason": "string"}, ...]}
Output: {"optimized_workflow_steps": [{"step_name": "string", "description": "string", "automation_suggestion": "string"}, ...], "estimated_improvement": "string"}
Execution harness: Python
- Task 3
Input: {"tool_invocation_request": {"tool_name": "string", "parameters": {"key": "value"}}}
Output: {"tool_response": "string", "success": "boolean"}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.