A day in my life as a tokenmaxxer 😎
3:21

A day in my life as a tokenmaxxer 😎

What's AI by Louis-François Bouchard 03.05.2026 212 просмотров 2 лайков

Machine-readable: Markdown · JSON API · Site index

Поделиться Telegram VK Бот
Транскрипт Скачать .md
Анализ с AI

Оглавление (1 сегментов)

Segment 1 (00:00 - 03:00)

So, you're being called one of the first token maxers in history. Can you explain what a normal day in your life looks like? Correct. So, basically, I wake up at around like 6:00 a. m. and before I even open my eyes, I ask Claude Code to summarize my dreams from the data I have from my Amazfit and my Garmin. Then, I ask ChatGPT to critique Claude's summary. Then, I ask Codex to check if the dream had any security vulnerabilities of the government trying to hack me. Security vulnerabilities in your dreams? Yeah, exactly. My Garmin showed I had elevated stress at 3:12 a. m. and my Amazfit actually detected light sleep. So, obviously, Claude concluded I was being targeted by a low-latency government prompt injection. Codex confirmed the dream was not secure, so I have to regenerate the night with stricter permissions, but I'm still working on that one. Right. And what happens after you wake up? Right. So, I have about 10 agents running in parallel. One agent checks my calendar, one checks emails, one decides if I should drink water, one opens Deep Research to compare water against coffee based on my body readiness and HRV. Then, I have a PM agent supervising those agents and another PM agent supervising the PM agent and myself because I don't want to be too hands-on. Do you actually read what they send you? No, of course not. I approved all last December and it's been running since. That's the whole point. Reading is basically manual prompting and why would I do that? It's like so 2025. And I heard your home is connected to your Claude bots? Yeah, my fridge is connected to Open Claude. So, if I'm out of milk, for example, it pings my PM, Claude creates a shopping plan, Gemini compares the milk brand, Grok checks all daily tweets about controversial opinions on cow or oat milk, and Codex makes the order to be delivered with a computer use access. Do you ever go to the store? I asked my agent to order it, then another agent to check if the first agent hallucinated the milk. I don't really go out anymore, in fact. I just waste too much token potential if I didn't look at the agents. How many tokens do you use per day? Honestly, yesterday was a bit of a light day. That's my bad. Only about 4. 5 million tokens before lunch. Shipping three new Stripe first coaching products where I teach people to spend more tokens. Wait, before lunch? Yeah, but lunch is automated now. I have one agent checking my macros, one agent checking if eating is aligned with my Q2 goals, and one agent ordering food. Then, Codex reviews the order for security again, Claude rewrites it in a more scalable way, and Gemini checks if the sandwich has product market fit aligned with my stress levels. So, what would you say to people who think this is wasteful? I would say the real waste is definitely paying for these models and not using them for full potential. You think I really bought eight Mac minis to ask one question at a time like a farmer? Yeah, manual prompting. It's very 2024. If you end the day with unused context windows, unused agents, and unused reasoning tokens, that's just value left on the table. It's pure waste.

Другие видео автора — What's AI by Louis-François Bouchard

Ctrl+V

Экстракт Знаний в Telegram

Экстракты и дистилляты из лучших YouTube-каналов — сразу после публикации.

Подписаться

Дайджест Экстрактов

Лучшие методички за неделю — каждый понедельник