Strategic Market Intel
In light of the intense competition among AI giants like OpenAI and Anthropic, this challenge focuses on building a dynamic competitive intelligence system. Participants will use Mastra AI to orchestrate a multi-agent team that gathers, analyzes, and synthesizes market data to provide strategic insights. The system will leverage Claude 4 Opus for sophisticated analysis, Gemini 3 Flash for rapid data processing, and integrate with various data sources using Lyzr, while utilizing Upstage for specialized document understanding. The goal is to generate actionable strategic recommendations for a hypothetical AI startup.
What you are building
The core problem, expected build, and operating context for this challenge.
Develop a Mastra AI multi-agent system with Claude 4 Opus and Gemini 3 Flash for competitive market intelligence and strategic recommendation generation.
How work is evaluated
The solution will be evaluated on the accuracy and depth of its strategic recommendations, the efficiency of its data processing pipeline, and the robustness of the multi-agent orchestration. The comprehensiveness of the market analysis and the clarity of generated reports will be key.
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.
Report Structure Adherence
The generated report must contain all specified sections in a logical order.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Agent Workflow Completion
All defined agents in the Mastra AI workflow must complete their tasks without critical errors.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Strategic Insight Score
Expert evaluation of the report's depth, originality, and actionability of recommendations (0-100). • target: 80 • range: 0-100
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Data Extraction Accuracy
F1-score for entity extraction from unstructured text (0-1). • target: 0.9 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Master Mastra AI's agent definition, memory management, and workflow orchestration capabilities to build a robust competitive intelligence system.
Implement agents with Claude 4 Opus to conduct deep strategic analysis, synthesize complex market trends, and formulate actionable business recommendations.
Integrate Gemini 3 Flash for high-speed processing of large volumes of market data, including summarizing news articles, financial reports, and social media feeds.
Design specialized agents that utilize Upstage's document AI capabilities to extract structured information from unstructured text, such as competitor press releases or research papers.
Build data ingestion tools using Lyzr's low-code integration platform to connect Mastra AI agents with various external APIs and web data sources (e.g., news APIs, financial data services).
Develop A2A communication protocols within Mastra AI to enable seamless collaboration between 'Market Researcher', 'Data Analyst', and 'Strategy Formulator' agents.
Explore patterns for deploying and managing Mastra AI agents in a production environment, considering aspects often handled by platforms like Sema4.ai (e.g., scalability, monitoring).
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The solution will be evaluated on the accuracy and depth of its strategic recommendations, the efficiency of its data processing pipeline, and the robustness of the multi-agent orchestration. The comprehensiveness of...
Challenge input
JSON object with 'market_segment' string (e.g., 'Enterprise LLM Providers') and 'competitor_list' array of strings.
Mastra AI
TypeScript agent framework.
Lyzr
Enterprise AI agent builder
Claude 4 Opus
Policy Serving in the agent workflow.
Evaluated output
Markdown file containing the strategic report, including SWOT analysis, market share estimates, and recommendations.
- The generated report must contain all specified sections in a logical order.
- All defined agents in the Mastra AI workflow must complete their tasks without critical errors.
- Expert evaluation of the report's depth, originality, and actionability of recommendations (0-100). • targe...
- Strategic Insight Score target: 80
- 2 public reference cases
- TypeScript 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
- Mastra AI · Required
- Lyzr · Optional
- Claude 4 Opus · Optional
Evaluation contract
- Report Structure Adherence · Weight 1
- Agent Workflow Completion · Weight 1
- Strategic Insight Score · Weight 1
- Data Extraction Accuracy · Weight 1
Metrics
- Strategic Insight Score · Target 80
- Data Extraction Accuracy · Target 0.9
Reference cases
- 2 public reference cases
Workflow tasks
- Task 1
Input: JSON object with 'market_segment' string (e.g., 'Enterprise LLM Providers') and 'competitor_list' array of strings.
Output: Markdown file containing the strategic report, including SWOT analysis, market share estimates, and recommendations.
Execution harness: TypeScript
- Task 2
Input: Text content of a press release.
Output: JSON object with extracted entities.
Execution harness: TypeScript
Recipe freeze marker
No recipe freeze time is recorded.