AgentOp
Finally shipping this! Real LLMs running in your browser no install, no server, nothing leaves the tab. Build agents in Python, export as one HTML
VisitPublished on:
June 21, 2026
Pricing:

About AgentOp
Hey 👋
Quick story on why this exists. I got tired of three things: spinning up a server just to show someone a small AI agent, pasting my API key into random apps, and shipping my data off somewhere just to get a simple answer back.
AgentOp is me trying to fix all three. It runs open models (Llama, Qwen, Gemma, Phi…) right in your browser with WebGPU. No install, no server, no key needed. You write the agent's logic in Python, test it against a local model right there, then export the whole thing as a single HTML file. One file. You can email it, drop it on a USB stick, or open it offline, and it just runs on the other person's machine with nothing to set up.
The part I'm weirdly proud of: open your network tab while chatting with a local model and it's empty. Nothing leaves the page. And if you'd rather use GPT-4 or Claude, you can bring your own key — it gets encrypted in the browser and never hits my servers.
There's also a newer feature where you just describe the agent you want in plain English and it writes the whole template for you. It's not perfect yet, so honestly I'd love for you to try and break it.
It's free to start. I've been heads-down on this for a while and I'm sure there are rough edges, so any feedback (the blunt kind especially) genuinely means a lot. What would you build with it?
Similar to AgentOp
AI Chat, Chat AI, AI Detector & AI Checker. Chat with AI, detect AI writing, humanize text and create content powered by Chat GPT 5.5.
SOC 2 and ISO 27001 turned into a clear to-do list. AI prepares the first drafts, you approve every call, and the founders actually answer.
Track every marketing link, attribute every dollar, and run your affiliate program — under one login.
Distro is an AI Distribution Operator that helps B2B teams publish content, find buyer conversations, engage prospects, and turn social intent into pi