NVIDIA-AI-Blueprints/aiq

What it solves

AI-Q is a deployable research backend designed for governed, citation-backed research workflows. It solves the problem of generating high-quality, in-depth research reports and quick answers while allowing organizations to self-host the application boundary and connect their own models, data sources, and policy controls.

How it works

AI-Q uses a LangGraph-based state machine to orchestrate different research depths based on user intent:

• Orchestration Node: Classifies whether a request is a general greeting (meta) or a research task, and determines if it requires a "shallow" or "deep" research approach.
• Shallow Research: A fast, tool-augmented agent that provides quick, cited answers.
• Structured Deep Research: A multi-phase process involving advisory source routing, structured planning, concurrent researcher workers, and a dedicated writer to produce comprehensive reports.
• Knowledge Integration: Connects to various data sources including web search (Tavily, Exa, Nimble, You.com), academic papers (Serper, SerpAPI, SearchAPI), and enterprise knowledge layers (OpenSearch, Azure AI Search, LlamaIndex).
• Execution & Storage: Uses provider-neutral sandboxes for code execution and supports durable output storage in SQL or S3-compatible systems.

Who it’s for

• AI researchers and developers building or extending agentic research workflows.
• Enterprise teams requiring tool-augmented research with strict citation requirements.
• NeMo Agent Toolkit users looking for advanced multi-agent patterns.

Highlights

• Multi-depth Research: Offers both fast shallow research and comprehensive, structured deep research reports.
• Modular Workflow: Agents (orchestration, shallow, deep, clarifier) are composable and can run standalone.
• Governed Research: Includes NeMo Guardrails middleware for policy controls and narrow application-level encryption for async outputs.
• Extensible Data Sources: Supports a wide array of search providers and enterprise knowledge backends.
• Evaluation Harnesses: Built-in benchmarks (e.g., FreshQA, DeepResearch) to measure and iterate on quality.
• Deployment Flexibility: Provides assets for Docker Compose and Helm for Kubernetes deployments.