Multi-Agent News Synthesis
Develop an advanced multi-agent system designed to revolutionize news content creation for publishers. This system will autonomously gather information from diverse sources, perform in-depth analysis, detect potential biases, and generate structured content drafts tailored to specific publication styles (e.g., concise summaries for social media, detailed analyses for financial news). The challenge emphasizes leveraging modern generative AI for content production, while ensuring accuracy and mitigating bias through robust agent collaboration and validation steps. The core of the solution will involve orchestrating a team of specialized agents, each responsible for distinct tasks such as research, factual verification, stylistic adaptation, and final content generation. The system should be able to process real-time information, adapt to evolving news narratives, and produce high-quality, relevant content that adheres to editorial guidelines.
What you are building
The core problem, expected build, and operating context for this challenge.
Orchestrate a multi-agent system with Fetch.ai and Gemma 2 for automated, bias-aware news content generation and synthesis.
How work is evaluated
The evaluation will assess the system's ability to autonomously generate coherent, factual, and stylistically appropriate news content, as well as its effectiveness in bias detection and secure data handling. Submissions will be tested against a set of real-world news events and publisher guidelines.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Master Fetch.ai's agent framework to define and orchestrate a team of specialized agents (e.g., Researcher, Verifier, Synthesizer, Editor).
Implement information gathering and preliminary summarization using Google's Gemma 2 Large Language Model.
Design and integrate content validation workflows using Guardrails AI to detect factual inaccuracies, logical inconsistencies, and potential biases in generated drafts.
Build secure API integration points using Anon authentication for accessing proprietary news feeds or internal knowledge bases.
Deploy and manage Gemma 2 and other foundational models for inference on Amazon Bedrock, ensuring scalability and cost-efficiency.
Orchestrate a dynamic content generation pipeline where agents collaboratively refine outputs based on real-time feedback and predefined editorial guidelines.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The evaluation will assess the system's ability to autonomously generate coherent, factual, and stylistically appropriate news content, as well as its effectiveness in bias detection and secure data handling. Submissi...
Challenge input
{ "event_summary": "string", "target_publication_style": "string" }
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{ "article_draft": "string", "summary": "string" }
- Verify that Fetch.ai agents can communicate and coordinate effectively to complete the content generation w...
- Ensure secure authentication is used for external API calls, demonstrating Anon integration.
- ContentFactualAccuracy 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
- Verify that Fetch.ai agents can communicate and coordinate effectively to complete the content generation workflow.
- Ensure secure authentication is used for external API calls, demonstrating Anon integration.
Metrics
- ContentFactualAccuracy · Target 0.9
- BiasDetectionRecall · Target 0.8
- StylisticAdherenceScore · Target 0.75
Workflow tasks
- Task 1
Input: { "event_summary": "string", "target_publication_style": "string" }
Output: { "article_draft": "string", "summary": "string" }
Execution harness: Python
- Task 2
Input: { "generated_content": "string" }
Output: { "bias_flags": [{"bias_type": "string", "severity": "enum(low, medium, high)", "mitigation_suggestion": "string"}] }
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.