Where I'm pointed next.
“Currently exploring how AI can merge with frontend to create delightful experiences.” That line from my profile is the work — here's what it looks like.
The UI is the prompt.
Every click, scroll and dwell is signal. A good AI-native interface uses that signal to compress steps — not to bolt a chatbot onto a finished screen. I'm building toward product surfaces where intelligence is ambient, not a separate tab.
Claude, MCP and n8n are the trio I'm betting on — reasoning, tools, and orchestration. The future is agentic, and the front-end is the surface where it lands.— Hemant
Claude-powered product surfaces
Using Claude as the reasoning layer for AI-assisted UX — smart defaults, inline suggestions, adaptive empty states. Not a chat tab; an ambient layer.
MCP servers as first-class tools
Building custom Model Context Protocol servers that expose real tools and data to LLMs — the foundation of every reliable agent I'm shipping.
Agentic workflows with n8n
n8n is my agent orchestrator — chaining LLM calls, tools and human-in-the-loop steps into workflows that actually run. Hosted in my homelab.