If I broke my career into chapters, there would probably be three. Banking, tech, and now international development. For the first two, I relied pretty heavily on personal assistants. Not The Devil Wears Prada kind. Mine were remote, mostly based in the Philippines, and cost me around $700 a month.
In banking, when I was on the trading desk by 6:30 a.m. and routinely working 12-hour days, that felt like an absolute bargain. Groceries, gifts, vacation ideas, reservations, random research, all the little things I technically could do myself but had neither the time nor brain capacity to deal with. The model worked so well that I kept it when I was running my startup, when the hours somehow became even more ridiculous.
The Philippines also happened to be perfect for this in the 2010s. No real language barrier, great talent, and the time difference meant that when my day ended, theirs was beginning. I could send a list before bed and wake up with half my life mysteriously handled. For an overworked finance or tech person, it was kinda magical.
But over time, it became more than task delegation. When you work with someone for years, even mostly through email, WhatsApp, and a weekly Skype call, they start accumulating context about you. They know how you think, what you like, what you tend to forget, and which ideas you will probably abandon three days later. I’d bounce ideas off them, ask for opinions, and sometimes just use them as another perspective on whatever was occupying my brain.
There was obviously an imbalance. I was the client, and even when I asked for brutally honest feedback, I understood there was probably some sugar coating involved. Their job, after all, was to make my life easier, not tell me I was an idiot every Tuesday morning.
Then I moved into government and something strange happened. For the first time in about 15 years, I had something resembling a normal schedule. I also traded a banking and tech salary for a government one, so the economics changed pretty quickly. I stopped having an assistant and, somewhat embarrassingly, really missed it.
Then generative AI got good. Not early ChatGPT good, but the newer models that can reason, remember context, run workflows, and actually be useful beyond telling you how many teaspoons are in a tablespoon.
Suddenly, the old relationship came back in a different form. I have AI agents that curate news for me, track concerts nearby, help plan trips, monitor parts of my finances, and handle a bunch of things my human assistants used to do. I probably spend $100 to $150 a month on tokens across different tools, which is significantly cheaper than $700 a month and comes with fewer scheduling conflicts.
The funny part is how natural the transition felt.
For a lot of people, having an ongoing relationship with an AI assistant still feels strange. For me, I realized I had been practicing for it for more than a decade. My relationship with human assistants was already mostly text-based, asynchronous, and built through hundreds of little exchanges that gradually created context and trust. The interface barely changed. What changed was who, or maybe what, was sitting on the other side.
Of course there are tradeoffs. With a human assistant, I was trusting someone halfway around the world with pieces of my life. With AI, I’m trusting companies and their systems with many of those same pieces. Different risks, same basic bargain. Convenience requires some level of trust, whether the assistant is in Manila or sitting inside a data center somewhere.
If the next chapter of my career brings back the 12-hour days, I’m pretty sure I’ll have a personal assistant again.
I’m just not sure that assistant will be human.
Roll credits.