Many years ago, I was invited to an adventure. My colleague from Facebook was assembling a little skunk works crew for a mission at Apple’s headquarters. We were to build the first prototype of the Facebook app for Apple TV. It was 2014.
Our team of 5 had React and GraphQL legends. I felt very fortunate and very excited.
We got to work. But instead of building the actual app, we jumped into building the infrastructure. We ported React to work with then not-yet-announced TVML, used the limited APIs of their JS flavor to talk to our GraphQL servers, set up auth, CI and many other developer experience basics.
After a day of this, no progress was made on the actual app. The Apple guy was neither impressed nor disappointed – it was normal for big companies to bring their engineers to Apple’s HQ for “partnership” where nothing useful would come up with that.
But we went home excited for the next day. We knew what leverage we built for ourselves. I kept hacking on the CI/CD pipeline well into the night, reducing the overhead down to a few seconds.
The next day started just like the previous one. After coffees and small talk, the Apple dived into reading the news, expecting nothing significant.
But the next time he looked at the shared monitor connected to the Apple TV, something very strange was happening. We had the first version of the app on the screen. And then the second. And the third. In less than an hour we iterated on several different concepts, the UI changing almost instantly and showed real production data. We had our friends list, photos and videos show up on the TV, we could rearrange everything.
The Apple guy was blown away. From his perspective we made almost no progress at first but then everything happened all at once.
From my perspective, I was having the time of my life. The system translated our intentions into the real product at the insane speed. Any idea you had for the experience, you could make it and see it on the real device right away.
If you look at what’s happening with AI software development early 2026, it does seem very familiar. A handful of engineers investing in their software factories, which from outside looks unnecessarily complicated and are yet to produce anything of value. But don’t dismiss them yet. Pay attention.
Slow at first, but then suddenly all at once.
P.S. That speed was intoxicating, but in the end it didn’t matter. The Facebook Apple TV app never shipped. The leadership poured all resources into mobile and didn’t have time for the side quests.