About two weeks ago, on a Saturday, I sat down to try AI assistants. I wanted to see how far they actually go, beyond the demos. And I had another reason, laziness: to free up some time. I collaborate in a Skool community about AI and that takes extra work, so I wanted to see if I could hand part of it off and, while I was at it, try to pick the podcast back up.
I started with a single assistant, Leonardo, who acts as coordinator, something like a chief of staff. On October 5 I put a small team together around him, which I call the Dream Team: Hubble researches, Cicero builds, Feynman critiques, and Galileo verifies. They share a few common rules, and each one lets the next know when they’re done, like a tiny assembly line. The next day I organized it with a coordinator and specialists by area, and I added Franklin, who handles my email without sending anything, and Ada, who helps me with the Skool community. Then came Socrates, for my philosophical chats, and Sagan, who helps me with content. I chose this structure because it was the model Grok Bot recommended and because, after spending some time studying how these assistants work, it seems to me the right orchestrator pattern.
Each one takes on a historical character, with their own tone and portrait (yes, I find it funny to talk about email with Franklin). But the idea underneath is serious: each one has a clear job, I only talk to Leonardo, and the rest work underneath. And there’s a rule that isn’t negotiable: nothing gets published or sent without my approval. When something goes out, it carries the virtual assistant’s signature, never mine.
The most routine part is Hubble’s: every morning he prepares the news on AI, technology, business, and the economy, and all I have to do is read it.
Another very ordinary example. One Monday I left late for Segovia and La Granja de San Ildefonso, and Leonardo redid the plan on the fly: he gave me the address for the GPS and where to park, looked up places that were open for a meal at an odd hour, gave me a summary of the palace right as I was walking in, and recommended where to get a coffee.
On Skool, Ada prepares the messages and I read and approve them before they’re posted. For example, this welcome to someone coming back to the community: “Welcome back! Good to see you picking it up again. Come by the AI sessions to catch up at your own pace. Any questions, we’re here.” Or support: recently a member wrote to me privately because an AI tool was throwing an error when it couldn’t find a text file they had pasted earlier. Ada drafted the reply, which I reviewed before sending: it explained that the tool no longer had that file available, that they should attach it again or paste the content directly into the chat, and that if it kept failing, they should send us screenshots.
Mind you, this isn’t new. OpenClaw has existed since February 2026, and there are others like Hermes. The difference is that here the technical complexity disappears, and that seems wonderful to me. The key point for me: the desktop apps for ChatGPT or Claude depend on my computer being on, while this lives in the cloud and keeps working even if mine is off.
What I haven’t tried are the assistants from ChatGPT and from Meta: Dots, ChatGPT’s, isn’t available in the European Union, and Muse, Meta’s, hasn’t even arrived. Thanks, European Union, for protecting me with absurd laws from the tools the rest of the world is already using.
Not everything is perfect. Sometimes I have to repeat instructions, voice recognition still isn’t very precise, and for a while now the Spanish voices sound like English speakers speaking Spanish, which is very noticeable and a bit annoying. Even so, I find all of them useful and I wouldn’t drop any, because the team is tailored to me. Training it does take some work at the start, but after that you only have to adjust it now and then.
For 30 USD a month I save hours I can spend learning or on other things. In my opinion, it’s worth it.
Usually, the person talking about AI is an influencer or a business owner. I want to speak to anyone who’s curious, from ordinary day-to-day use, so this is useful to them and takes those banal tasks that steal our time off their plate. So I’d love to know: would you be willing to spend that money to try it? And if you did, what bots would you create?