How I connected AI to my tools
The MCPs I use every day, how they're installed and the risk of giving it access to everything.
You do it from ChatGPT's settings, in the apps section (they used to be called connectors): you choose Gmail, sign in with your Google account and accept…
You do it from ChatGPT's settings, in the apps section (they used to be called connectors): you choose Gmail, sign in with your Google account and accept the permissions (checked in September 2026). From then on you can ask it to find emails, summarise a thread or prepare a draft for you. Whether the option shows up depends on your plan and your country, so the first thing is to check whether it's available on your account.
If it doesn't show up, there are other routes: automation tools that act as a bridge between your email and the AI, or the good old method of copying and pasting the email into the chat. What usually confuses people is thinking that, once connected, ChatGPT watches your inbox and replies on its own. It doesn't: it works when you ask it to. And it's worth thinking twice before connecting your work email, because there's client and third-party data in there, not just yours. Check what permissions you grant and how to revoke them if you change your mind.
I work mainly with Claude, where the same thing is done with MCP, but the lesson applies to both: connecting is quick, and what takes time is deciding what the AI can read and what it can do on its own. I explain it in "How I connected artificial intelligence to my tools", where I cover what connecting AI to your tools actually means and the risks of giving it access to everything.
The MCPs I use every day, how they're installed and the risk of giving it access to everything.
One email a week with what was verified, what was refuted and what is still on the bench. No news roundups.