This blueprint demonstrates how to build an AI agent that automates blog post creation by leveraging LlamaIndex's document processing and agentic framework capabilities alongside NVIDIA Llama 3.3 70B LLM NIM and NeMo Retriever E5 embedding NIM. The agent handles the entire writing process - from research and outlining to drafting and editing - to produce high-quality, well-researched blog content.

The system provides an automated workflow that orchestrates the entire content creation process - from initial research and outlining to drafting and quality assurance - while maintaining high standards of accuracy and comprehensiveness through the integration of LlamaIndex's document processing with NVIDIA's powerful language and retrieval models. Developers can use this flexible reference blueprint to update an existing LlamaIndex solution with NVIDIA AI, create a new blog creation agent, or apply it to a different use case that includes LlamaIndex and NVIDIA.

Architecture Diagram

Key Features

Ingestion Phase

1. Document parsing: Use LlamaParse to convert PDFs into Markdown.
2. Document ingestion: Employ NeMo Retriever E5 embedding NIM to ingest the documents into a vector database.
3. Query engine instantiation: Leverage LlamaIndex to create a query engine that can answer questions about the documents.

Query Phase

1. Research Request: User provides a set of tools (e.g., a query engine with data about San Francisco's budget) and a content request (e.g., a question for a blog post).
2. Outline Generation Agent: Deploys an agent to structure the blog post into an actionable outline.
3. Question Generation Agent: Generates a list of questions necessary to address the outline effectively, and breaks the questions into discrete units that can be answered concurrently, using available tools for data collection.
4. Content Generation Agent: A writer agent synthesizes the gathered answers into a cohesive blog post.
5. Critic Agent: reviews the content for accuracy, coherence, and completeness, determining if revisions are necessary.
6. Iterative Refinement: If improvements are needed, the workflow repeats by generating additional questions and gathering more information until the desired quality is reached.

Minimum System Requirements

The solution leverages NVIDIA's cloud-based API Catalog endpoints, eliminating the need for local GPU hardware. All model inference is performed on NVIDIA's cloud infrastructure.

Software used in this blueprint

NIM microservices

• Llama 3.3 70B NVIDIA NIM
• NeMo Retriever E5 embedding NIM

3rd-Party Technologies

• LlamaIndex

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility, and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure the models meet requirements for the relevant industry and use case and address unforeseen product misuse. For more detailed information on ethical considerations for the models, please see the Model Card++ Explainability, Bias, Safety & Security, and Privacy Subcards. Please report security vulnerabilities or NVIDIA AI concerns here.

License

Use of the models in this blueprint is governed by the NVIDIA AI Foundation Models Community License.

Terms of Use

GOVERNING TERMS: The blueprint is governed by the NVIDIA Agreements | Enterprise Software | NVIDIA Software License Agreement and NVIDIA Agreements | Enterprise Software | Product Specific Terms for AI Product.

Meta Llama 3.3 70B

GOVERNING TERMS: The NIM container is governed by the NVIDIA Software License Agreement and the Product Specific Terms for AI Products;

Use of this model is governed by the NVIDIA AI Foundation Models Community License Agreement. ADDITIONAL INFORMATION: Llama 3.3 Community License Agreement, Built with Llama.