Since I subscribed to Claude Max (about half a year ago, I think), I have been much more productive. Some might laugh and say AI just gives you a false sense of productivity—that it does all the work while you sit in front of a screen doing nothing. But that isn’t true.

What I do with AI

I contributed A LOT to the design and decision-making process. I made sure to deeply understand the problems and the AI’s proposals, and in many instances, I had huge debates with it. Some even took hours and consumed 60-70% of the context window just to clarify things, evaluate tradeoffs, future-proof the system, and prepare the operations, specs, and plans. Thanks to AI, I was able to offload the heavy lifting of typing and debugging (I am terrible at debugging anyway, as I am highly prone to confirmation bias) and focus on the Products, the Architectures, the Philosophies, the Human element, and the Impact (I believe I already mentioned this somewhere else). I always thought these were the interesting aspects that a senior should focus on, and up to now, I still believe that’s true (at least from my POV).

But it is exhausting…

How I have been doing with AI

Designing an architecture or a product, while it sounds grand, doesn’t actually consume that much continuous mental bandwidth. We usually need to pay full attention to it for a few hours, or even days (or weeks for enterprise work with multiple stakeholders), to do all the analysis and planning, make the decisions, and document them permanently so they hold up for months. Once things are seemingly perfect (given the specific context) and there is little to no scope drift, the high-level work stops. The rest is just the tedious work that AI handles mostly (if not completely) for us.

(open zoomable view) Because of this, there is an “urge” to be even more productive while waiting for the AI to do the tedious work. If you are a software engineer, you can probably predict what I did next. It is exactly like a CPU being stretched to its absolute limit.

(open zoomable view) In an attempt to maximize my output, I even tried working on parallel projects simultaneously—context-switching and answering questions for one project while the other project’s agent was generating a response.

The result

I achieved a lot in all regards. I built an AI Orchestrator, a personal Home Server with all the Platform features, and more. I learned a lot and solidified my knowledge base, especially in the Platform domain. My products are all fairly good and well-designed, too. However, the constant context-switching fried my brain. I was exhausted to the core and started experiencing bad symptoms: lack of sleep, loss of motivation, loss of focus, waking up in the middle of the night… And then I would turn to social media and doomscroll, which made things even worse.

What I am doing now

I downgraded to Claude Pro so I no longer feel pressured to be hyper-productive. Now, I spend my time resting with my family and friends, focusing on my day job instead of trying to build a legacy. I am not sure if my health will fully recover, but I sincerely hope that others will not fall into this same trap.