A mind for Noesis
With Noesis I am building a real mind for my agent system. I give it limbs, which are the plugins. I give it skills, abilities it has to remember all the time, so it never makes the same mistake twice and does not get lost in overly long reasoning. And I give it a memory. That way I can drive it with direct tasks, and it keeps learning what it has to do. This morning, through ChatGPT, I asked it to check all my email: it looked at what needed attention most and what did not. The more it knows about me, the more it learns.
The local watchdog
I added another 32 GB stick of RAM. It lets me run a local model, a quantized Qwen, as a small engine that is always on and acts as a watchdog. When the system gets stuck, does not know how to move on or starts looping, the small system spots the gear that is not turning, adds some oil and frees it. It is like aircraft engines: to start the big engines you always need a small starter motor.
Base camps on Linear
Every day I update my project on Linear, connected to ChatGPT. In the evening I say what I want to do tomorrow and I get the task list: check Gmail, check WhatsApp, check whether something is wrong in the business. In the morning you are no longer going in blind. You also need a big goal: to climb Vesuvius, where I live, you do not do it in one day if you want to explore all of it. You set base camps: camp one, camp two, camp three. If it gets too much, you go down a bit, fix what was left behind and start again.
One agent for every task
If your agents just sit there doing nothing, you are wasting tokens and resources that are practically being given to you. You can set them up like this: one agent for email, one that checks on work, one for family communication. Even for your kids' homework: you send a photo and the agent checks whether it is done right and where the difficulties are. I want your agents working around the clock, taking the dirty work, the bureaucracy and the boring stuff off your plate. Even one extra hour a day to train, walk or stretch changes everything.
The morning question
You can start right now. Wake up and ask yourself: what is making me waste time? What is the repetitive thing I cannot stand anymore? That is where ingenuity starts, and that is where I tell you which agent to use. Write it in the comments under the video. I have used AI for three years, at least ten hours a day, and I still have bottlenecks and blind spots. AI literally saved my life: it is like a fish taken out of its bowl that realises it has spent its whole life in there. It puts you in a position to observe yourself, see where you go wrong and which metrics you are not meeting.
Video: Hai agenti AI e li lasci a non fare niente: io gli ho dato una mente (Noesis) (in Italian) · Italiano: versione italiana · Yesterday: The bottleneck is not AI, it is you · Community: Body Mind Systems on Skool.