Deployment

August 21, 2026 · View on GitHub

The AI-Q blueprint supports multiple deployment methods. Choose the one that best fits your environment and operational requirements.

Method	Best For	Prerequisites
Docker Compose	Local development, team demos, single-node deployments	Docker Engine, Docker Compose v2
Kubernetes (Helm)	Multi-node clusters, production	Kubernetes cluster, Helm v3.x
Manual (no containers)	Development and debugging	Python 3.11--3.13, system dependencies (refer to Installation)

Published Release Artifacts

AI-Q v2.2.0 is published on NVIDIA NGC. Use the exact versioned references below for release deployments.

Artifact	Type	Versioned reference
`aiq-agent`	Container image	`nvcr.io/nvidia/blueprint/aiq-agent:2.2.0`
`aiq-frontend`	Container image	`nvcr.io/nvidia/blueprint/aiq-frontend:2.2.0`
`aiq2-web`	Helm chart	`nvidia/blueprint/aiq2-web:2.2.0`

Architecture Overview

All containerized deployments run the same three services:

Backend (aiq-agent) -- FastAPI server with an embedded Dask scheduler and worker for background job processing.
Frontend (aiq-blueprint-ui) -- Next.js web UI that communicates with the backend API.
Database (postgres) -- PostgreSQL instance for async job storage, LangGraph checkpoints, and document summaries.

Deployment Guides

Docker Compose -- Full Docker Compose reference covering environment setup, the standard LlamaIndex stack, Foundational RAG (FRAG) integration, database configuration, and troubleshooting.
Kubernetes (Helm) -- Helm chart deployment for Kubernetes clusters, including NGC image pull secrets, configuration switching, FRAG integration, and troubleshooting.
Amazon OpenSearch Serverless -- EKS and OpenSearch Serverless deployment notes for the built-in OpenSearch knowledge backend.
Docker Build System -- Multi-stage Dockerfile architecture, build targets (dev vs. release), base images, and startup scripts (entrypoint.py and start_web.py).
Authentication -- Enable OAuth/OIDC sign-in, configure backend JWT validation, and use AIQ user tokens in tools and MCP pass-through integrations.
Async Job Content Encryption -- Configure encryption at rest for async final reports and selected artifact event content, including Vault Transit and static-key modes.
Observability -- NeMo Relay logging, ATOF traces, Phoenix OTEL export, redaction, and cost data.
Production Considerations -- Guidance on managed databases, horizontal scaling, security hardening, monitoring, and resource requirements.
OpenShell -- Optional policy-bound execution for generated deep-research code, including supported platforms, authenticated gateway ownership, policy/config pairing, deterministic live acceptance, and safe cleanup.

Quick Start

For the fastest path to a running stack:

# 1. Configure environment
cp deploy/.env.example deploy/.env
# Edit deploy/.env with your API keys

# 2. Start services
cd deploy/compose
docker compose --env-file ../.env -f docker-compose.yaml up -d --build

Open http://localhost:3000 to access the web UI.