RCARS — RHDP Content Advisory & Recommendation System

AI-powered content intelligence for the Red Hat Demo Platform. RCARS reads every lab and demo in the RHDP catalog, understands what each one teaches, and uses that understanding to help teams find the right content, detect duplicate material, and identify items that should be retired.

What It Does

• Recommendations — Ask a question in plain English ("what should we show at a Kubernetes conference?") and get a ranked list of catalog items with rationales
• Content Overlap Detection — Pairwise semantic comparison across the catalog to surface duplicate or near-duplicate labs
• Retirement Analysis — Scores every catalog item based on provisions, sales impact, cost, and age to identify retirement candidates
• Infrastructure Metadata — Extracts workload mappings, cloud providers, and platform details from AgnosticD v2 configurations
• Catalog Browse — Filterable view of all catalog items with curator controls for content management

Architecture

Four deployments on OpenShift: React frontend, FastAPI API, arq workers (scan + recommend), PostgreSQL with pgvector. LLM analysis via LiteMaaS (primary) with Vertex AI fallback. Nightly pipeline handles catalog refresh, stale detection, workload scanning, content similarity, and reporting sync.

Documentation

Full documentation is published at rhpds.github.io/rcars.

• Overview — What RCARS is and how it works
• Web Guide — Using the web interface