Deploy NVIDIA AI-Q Blueprint on Oracle Cloud Infrastructure
NVIDIA has unveiled the AI-Q Blueprint, an open-source reference designed for advanced AI agents capable of executing complex tasks. These agents have evolved significantly over the past two years, progressing from simple question-answering systems to sophisticated long-horizon agents that can plan multiple steps, delegate tasks to sub-agents, and maintain context throughout lengthy interactions. The AI-Q Blueprint is built on LangChain Deep Agents and the NVIDIA NeMo Agent Toolkit, providing users with the ability to generate quick responses or comprehensive research reports with cited sources.
This article outlines the steps to deploy AI-Q 2.0 on Oracle Cloud Infrastructure (OCI) using Terraform for resource creation and Helm for workload installation. The deployment process is tailored for developers and platform engineers who are comfortable with Kubernetes, Terraform, and shell commands, and prefer to run AI-Q on OCI rather than on local machines.
For Your Trading Desk
Sponsor Spike Success.
The deployment guide details how AI-Q’s multi-agent architecture integrates with OCI services, providing exact commands to provision, deploy, and access the AI-Q blueprint. The architecture includes components such as an intent router that directs user queries to the appropriate workflows, enhancing the system's efficiency.
The AI-Q Blueprint is designed for extensibility, allowing users to swap out various layers—models, tools, and sub-agents—through YAML configuration or the NeMo Agent Toolkit plugin system. This flexibility will be explored further in subsequent parts of the series.
To begin the deployment, users must configure Terraform variables by copying an example file and editing it with their tenancy details. The initial setup takes approximately 10 to 15 minutes, during which Terraform creates essential resources like the Virtual Cloud Network (VCN), OKE cluster, Load Balancer, and a secure Vault for API keys.
Once the infrastructure is set up, users install AI-Q from the NGC Helm chart, which simplifies the process by eliminating the need for local builds. The installation involves configuring kubectl for the OKE cluster, creating necessary secrets, and pulling the Helm chart from NGC. After installation, users can verify the deployment by checking the status of the AI-Q pods and accessing the AI-Q frontend via a web browser.
With the deployment complete, users can test the AI-Q system by asking simple or complex questions to confirm its functionality. The entire setup can be easily dismantled with a single command, ensuring a clean removal of resources when no longer needed. Users are encouraged to clone the AI-Q repository and share their solutions on the NVIDIA Developer Forum, contributing to the community's knowledge base.




