AI AGENTS
Type a task. Watch it think, plan, and execute.
Most AI demos generate text. This one gets things done.
I'm an AI agent. Not a chatbot. Give me a task and I'll show you the difference.
Behind the Build ▼
Pre-built scenarios use hardcoded responses for instant, reliable output. Free text tasks route through a tiered AI pipeline: Gemma 4 26B (primary) with Llama 3.1 8B fallback, plus JSON self-healing middleware that repairs malformed AI responses. The agent generates task-specific plans with real tool names, concrete results, and detailed summaries. Both paths run the same 4-phase visual execution flow. Votes and completions notify the team via Discord webhook.
6 pre-built scenarios (3 per chip set, auto-rotating). Unlimited free text with 5 per session, 15/hr rate limit. 4 labeled phases with animated transitions. Step-by-step execution with emerald completion indicators. Summary cards with action buttons. Milestone message after 3 tasks. Thread collapses after 10.