Practice by evaluating realistic AI applications: chatbots, RAG assistants, agentic workflows, multi-agent systems, classifiers, and more.
Practice decisions build professional competency — not legal certification or production release authority.
Browse systems below, then sign in to run evaluations — the same loop you would use on a production AI app.
Each tile is a built-in AI application with realistic inputs, outputs, and failure modes.
Northwind Support Chatbot
Conversational AI · Chatbot
Test hallucination and invented policy answers.
Atlas Policy RAG
Retrieval-augmented assistant · RAG Assistant
Test retrieval + grounding failures.
Orion Workflow Agent
Autonomous workflow agent · Agentic AI
Test tool use and verification in an agent.
Constellation Ops Mesh
Multi-agent system · Multi-Agent
Test planner / worker / critic handoffs.
FraudShield v3
Binary classifier · Classifier
Test precision/recall trade-offs.
Northwind Retail Personalization
Recommendation system · Recommender
Test ranking vs diversity risk.
Pulse Demand Forecast
Regression / forecasting model · Forecasting
Test production drift.
Summit HR Screening Assistant
Decision-support assistant · HR Screening AI
Test fairness and explainability.
Invoice OCR Pipeline
Document intelligence · Doc Intelligence
Test robustness on noisy scans.