Collaborative Technical Research Crew
Design and implement a multi-agent system using CrewAI to collaboratively research and synthesize complex technical topics, inspired by detailed historical accounts like RISC-V development or DeepMind documentaries. The crew will consist of specialized agents (e.g., Researcher, Analyst, Technical Writer) working together. They will leverage Gemini 2.5 Pro via Vertex AI for deep understanding and content generation, utilizing ChromaDB as a shared knowledge base for storing research findings and contextual embeddings. Prefect will orchestrate the end-to-end research workflow, managing task dependencies and ensuring robust execution of the multi-agent collaboration.
What you are building
The core problem, expected build, and operating context for this challenge.
Orchestrate a multi-agent research crew using CrewAI, Gemini 2.5 Pro, and ChromaDB with Prefect for complex technical topic synthesis.
How work is evaluated
The CrewAI system will be evaluated on the quality, factual accuracy, completeness, and coherence of the generated technical research report. Efficiency of collaboration and proper tool utilization will also be assessed.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Master CrewAI for defining agent roles, goals, and tasks, orchestrating complex, sequential, and parallel workflows among collaborating agents.
Integrate `Gemini 2.5 Pro` as the core LLM for CrewAI agents via Google Cloud's Vertex AI, understanding authentication and API usage.
Implement a shared knowledge base for your CrewAI agents using ChromaDB, enabling agents to store, retrieve, and share research findings efficiently.
Design and deploy a Prefect flow to manage the lifecycle of your CrewAI system, including triggering research tasks and monitoring progress.
Develop custom tools for CrewAI agents (e.g., a 'Web Search Tool', 'Document Analyzer Tool', 'Report Generator Tool') to enhance their capabilities.
Apply advanced prompt engineering for `Gemini 2.5 Pro` to guide agents in complex reasoning, summarization, and technical writing.
Build an evaluation framework to assess the factual accuracy, completeness, and coherence of the collaboratively generated research reports.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The CrewAI system will be evaluated on the quality, factual accuracy, completeness, and coherence of the generated technical research report. Efficiency of collaboration and proper tool utilization will also be assessed.
Challenge input
String: 'Research topic'
Agent execution
No agent tool is listed for this challenge.
Evaluated output
String: 'Markdown formatted research report including sections like Introduction, Key Concepts, Historical Context, Current State, Future Outlook,...
- Checks if the generated report follows a basic markdown structure with expected headings.
- Checks if the fact-checking output matches the expected JSON format.
- Report Factual Accuracy target: 0.9
- 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
- Checks if the generated report follows a basic markdown structure with expected headings.
- Checks if the fact-checking output matches the expected JSON format.
Metrics
- Report_Factual_Accuracy · Target 0.9
- Report_Completeness · Target 0.85
- Report_Coherence · Target 0.8
Workflow tasks
- Task 1
Input: String: 'Research topic'
Output: String: 'Markdown formatted research report including sections like Introduction, Key Concepts, Historical Context, Current State, Future Outlook, Conclusion.'
Execution harness: Python
- Task 2
Input: JSON: {'report_section': 'string', 'statements': ['string', ...]}
Output: JSON: {'report_section': 'string', 'verified_statements': [{'statement': 'string', 'is_accurate': 'boolean', 'source': 'string'}]}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.