Agent-first AI tools playground
Pitaraजादू का पिटारा — A magic box of AI tools
Learn, build, and evaluate agentic AI — right in your browser. No sign-up. No server. Every page has a built-in AI copilot to help you.
Agentic Evaluator
Evaluate AI agent outputs with LLM-as-judge scoring. Define rubrics, run evaluations, and get visual report cards.
evaluateagentrubricllm-judge
LLM Chat
Chat with any AI model. Multi-thread conversations with full history. Test and compare models side-by-side.
chattestmodelscompare
Agent Loop Simulator
Visualize how AI agents process tool calls in the ReAct loop. Step through scenarios and watch costs accumulate in real-time.
learnagentvisualizationreact-loop
Runs in Your Browser
No server needed. Your data never leaves your machine.
Your Keys, Your Control
API keys are encrypted locally. We never see them.
AI Copilot Built-In
Every page has an AI assistant. Ask questions or let it drive.
An open sandbox for AI tools.