Just like the 3D printer promised to be the enabler of custom, homemade tools for the individual, I think AI is finally becoming that for software.
Like how I just made a tool for my job with Carrd, something inspired by the friction of replying the same customer support replies over and over. So I made a canned responses tool that I can add preset answers quickly, on the fly. Now, instead of searching, scrolling through a word doc, i just scan the app, and one-click to copy a chunk of text.
All done in 1-2h using AI.
And this isn’t the first time I’m doing this.
First it was ProfilePicture.app, a tool to make my own LinkedIn-inspired profile photos.
Then it’s GitChart to one-click download my Github charts for sharing.
Now, a canned responses tool.
And with AI agents on the horizon, whole workflows will be possible, not just one-off tasks.
Imagine the kind of throw-away apps you can build for yourself over a weekend. To enrich your own life, your family’s or your community’s.
I love how @karpathy describes it - “vibe coding”:
There’s a new kind of coding I call “vibe coding”, where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It’s possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like “decrease the padding on the sidebar by half” because I’m too lazy to find it. I “Accept All” always, I don’t read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I’d have to really read through it for a while. Sometimes the LLMs can’t fix a bug so I just work around it or ask for random changes until it goes away. It’s not too bad for throwaway weekend projects, but still quite amusing. I’m building a project or webapp, but it’s not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works. – @karpathy
Imagine this:
It’s the year 2050. You wake up to the sound of your AI alarm, calculating the sleep cycles to the minutes, waking you up at the right time.
The first thing you check is the outputs from the army of AI agents that were coding and reasoning about through the night for your latest startup idea. You see it’s already done, coded out in full stack from frontend to backend. Ready for you to just greenlight it.
YOLO, you think. And hit the glowing green button to launch. What’s the worse that can happen? Your army of AI agents can continue to fix bugs and iterate on it.
You are just washed up and making coffee when you get a notification that your marketing plans are ready. A full report on how to SEO, keyword difficulty, search volume, social media channels to sign up for, influencers to work with, etc. The full works. Your AI agent proceeds to brief you in the voice of Scarlet Johansson, how to get to the top of SERPs. You approve and ask it (Her) to provide updates every week.
A Stripe notification comes in. $10,000 MRR and climbing.
Things are going gooood.
And you’ve not even finished your morning coffee yet.
—
We truly are living in wild times.
It’s just not evenly distributed.
Yet.