Intelligent Document & Spreadsheet Generation with Gemini 2.5 Pro & CrewAI
Develop an advanced multi-agent system that leverages Gemini 2.5 Pro's Deep Think capabilities and CrewAI's orchestration to autonomously generate complex business documents and spreadsheets. This challenge focuses on automating comprehensive reports, financial analyses, or strategic plans, requiring agents to perform data extraction, synthesis, and structured content creation. Participants will design and implement a team of specialized agents capable of interpreting high-level requests, gathering necessary data through custom tools, and iteratively refining generated content. The system must demonstrate 'extended thinking' by pausing for clarification or executing multi-step reasoning processes to ensure accuracy and coherence, mirroring real-world business analyst workflows.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a multi-agent system using CrewAI and Gemini 2.5 Pro for automated business document and spreadsheet generation with extended thinking.
How work is evaluated
The solution will be evaluated on its ability to autonomously generate a detailed Q1 Financial Performance Report and a subsequent Marketing Campaign Plan based on simulated financial data and a high-level business brief. Accuracy, structure, and adherence to specific formatting requirements will be key.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Implement custom tools for agents to interact with document and spreadsheet APIs.
Design and orchestrate role-based agent teams for complex document creation.
Integrate Gemini 2.5 Pro's Deep Think mode for advanced data analysis and content synthesis.
Apply extended thinking patterns to iteratively refine agent outputs based on contextual feedback.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The solution will be evaluated on its ability to autonomously generate a detailed Q1 Financial Performance Report and a subsequent Marketing Campaign Plan based on simulated financial data and a high-level business br...
Challenge input
{'business_brief': 'string', 'mock_financial_data': {'Q1_Sales': 'float', ...}}
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{'report_title': 'string', 'sections': [{'heading': 'string', 'content': 'string', 'data_tables': [{'table_name': 'string', 'data': [['header1', 'h...
- Verify key financial figures and calculations in the generated report match the input data.
- Assess if the marketing plan strategies are relevant to the product and market trends provided.
- Verify agent logs show evidence of multi-step reasoning, tool usage, and iterative refinement.
- ContentCompleteness target: 90
- 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
- Verify key financial figures and calculations in the generated report match the input data.
- Assess if the marketing plan strategies are relevant to the product and market trends provided.
- Verify agent logs show evidence of multi-step reasoning, tool usage, and iterative refinement.
- Ensure the generated reports adhere to the specified JSON output format, indicating proper structured content generation.
Metrics
- ContentCompleteness · Target 90
- LogicalCoherence · Target 4
- ToolUseEfficiency · Target 5
Workflow tasks
- Task 1
Input: {'business_brief': 'string', 'mock_financial_data': {'Q1_Sales': 'float', ...}}
Output: {'report_title': 'string', 'sections': [{'heading': 'string', 'content': 'string', 'data_tables': [{'table_name': 'string', 'data': [['header1', 'header2'], ['row1_val1', 'row1_val2']]}]}]}
Execution harness: Python
- Task 2
Input: {'product_brief': 'string', 'market_trends': 'list of strings', 'q1_report_summary': 'string'}
Output: {'plan_title': 'string', 'sections': [{'heading': 'string', 'content': 'string', 'strategies': ['string']}]}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.