With the option to connect to an MCP Server in the OpenAI Dashboard, it’s becoming more useful to me than ChatGPT itself for building “agents”. I didn’t understand why AI Studio hasn’t implemented MCP as well.
I say this because no longer depending on a LangGraph helps A LOT to democratize more agentic use. I advise anyone to use it via the Dashboard and set up an MCP Server on Pipefy/n8n/Zapier. By the way, big kudos to n8n and how they abstract the use of LangGraph.
Oh, whenever I talk about this, people bring up Manus and Perplexity. I know very little about either, but as I’ve said in the past, I’m focused on understanding how the foundation models do it, and that’s why I implement things myself. Do I recommend it to those who don’t want to learn? No. Use something ready-made. Most likely conversational agents like ChatGPT/Gemini will get a way to use MCP, as Claude already has — Anthropic is another one I know little about.
Happy exploring in the world of MCPs -> which bring nothing new for those who already used a LangGraph, but which will help a lot to popularize tool use by agents.
Originally written in Portuguese and published on LinkedIn on July 30, 2025. I translated it into English with the help of AI (Claude): it let me bring my LinkedIn posts to people who don’t read Portuguese without rewriting each one from scratch. I reviewed the translation to keep my original meaning and tone.