Agentic research workflows on Red Hat AI Factory with NVIDIA

How do you move beyond basic single-prompt chatbots to handle multi-step enterprise research? Discover how Red Hat AI Factory with NVIDIA coordinates multi-agent research workflows on Red Hat OpenShift AI. Standard chatbots struggle with complex academic or business research that requires clarifying intent, gathering multi-source evidence, and synthesizing findings. This video demonstrates an enterprise research AI quickstart based on NVIDIA's AI-Q Blueprint, adapted to run on Red Hat AI Factory with NVIDIA. You will learn how an intent router evaluates incoming queries, delegating deeper investigations to specialized researcher subagents using LangChain and NVIDIA NeMo Agent Toolkit. Models are served through the vLLM (virtual large language model) inference stack, while private enterprise data is accessed via retrieval-augmented generation (RAG). Discover how platform operators track execution traces with OpenTelemetry and MLflow, monitor model endpoints in Grafana, and run automated assessments in Red Hat OpenShift AI workbenches. Timestamps 00:00 Chatbots vs. agentic research workflows 00:45 Architecture of NVIDIA AI-Q Blueprint on Red Hat AI 01:40 Data source configuration and intent routing 02:25 Deep research workflow execution and report generation 03:15 OpenShift AI infrastructure and vLLM model serving 03:50 Monitoring model serving metrics in Grafana 04:20 Tracing multi-agent execution with OpenTelemetry and MLflow 05:15 Automated workflow evaluation with MLflow in OpenShift AI workbenches 06:10 Key takeaways and getting started Resources 🤖 Try the Enterprise Research AI quickstart → https://docs.redhat.com/en/learn/ai-quickstarts/rh-rh-research ✨ Explore Red Hat AI portfolio offerings → https://www.redhat.com/en/products/ai #RedHat #AgenticAI #RedHatAI #NVIDIA #OpenShiftAI #MLflow #vLLM