As AI gets more and more capable, your limits of what's possible must expand as well 🏗️
But not necessarily your working time 🙌
AI can do more, and will do as much as you ask. It's again all back to you to decide: how much, how often, and how long 👇
How much
The answer is more. More tasks, bigger tasks, multiple tasks.
This requires rewiring your concept of what's possible 🤔
Personal example: this blog (sandromaglione.com).
Just 1 year ago, before AI, any feature added to the blog was a full new endeavour:
- Dealing with SEO, JSON-LD, metadata
- Implementing a new custom component for MDX parsing
- Adding social images generation for each post
- Adding a new search filter
Each of these tasks would have been a few days of work. As such, I needed to carefully consider which one was worth it, among other projects as well.
Today, one prompt implements all of them together 🙌
But I never stopped thinking about this until last week. My vision was still considering each task independently.
Part of your job as a software engineer is now to consider more ambitious tasks.
You can do any task now, but this does not mean that you should do all of them, all the time.
The other part of your job as a software engineer remains considering tradeoffs 👇
How often
Repetitive tasks are not a thing anymore (for the most part).
Large refactoring, migrations, redesigns. Those used to be big taboos, never to be mentioned. Now it's all a prompt away.
Again, being easy does not mean that you should always do it 💁🏼♂️
In code, there is always something that can be improved, refactored, migrated.
When anyone can do anything at all times, the leverage goes to whom can prioritize better 🙌
Software architecture matters more than ever. Your judgement on it matters more as well, ever than before 🫡
How long
Last key question.
Since in code there is always something to do, and AI can run at all times, what will you do? 🙌
For a considerable portion of people, the answer is always more (it seems).
Longer hours, more and more chats, trying to keep up with each AI.
The AI loop can quickly drain all your time 😬
Writing code used to be about intensity: how much it takes you to focus on a problem, and how long can you maintain that focus.
This lead to shorter "bouts" of focused work 🤔
Now it's about endurance: how long can you prompt until you lose track of what each AI is doing?
Since prompting is so easy, you can just send a message after another without noticing. And the reward loop (the uncertainty of it) pushes you to send more and more.
Read last's week newsletter for more details: How much is left to implement?
Mastery is about stacking new knowledge every day, with focused and deliberate work.
Prompts are shallow: you just say what, but never learn how. Don't fall into the trap 😬
This will piss you off but it's true
See you next 👋