
We think AI agents are new. Apple imagined them 38 years ago.
.jpg)


.jpg)

Watch this: https://www.youtube.com/watch?v=9bjve67p33E
This is a concept video Apple made in 1987.
Let that year sink in for a moment.
I know exactly what 1987 felt like, because I had already been teaching AI at university for several years by then. I was building programs on a Tandy TRS-80 with 4 kilobytes of memory. Four. And I had just completed my PhD thesis building one of the world's first multiprocessor computers. The idea of a consumer walking into a store and simply buying a computer was still a novelty. Harvey Norman wouldn't open its first computer superstore until 1993. In 1987, most people had never touched one.
And yet here is Apple depicting a professor having a natural conversation with an AI assistant that pulls research, filters journal articles, manages his calendar, coordinates with colleagues via video, and handles his day intelligently. On a device that looked like a tablet.
That is an AI agent. Nearly four decades ago.
Steve Jobs had actually left Apple two years before this was made, which makes it even more striking. This wasn't one visionary calling it. It was a team of people who could see clearly where computing was heading, long before the tools existed to get there.
That is the thing we keep getting wrong about this moment.
AI has not been invented. The concepts, the research, the architecture have been building for decades. I was teaching this material before most people knew the word. What is different now is not the idea. It is the ability to execute on it. The models are finally good enough. The compute is finally cheap enough. The interfaces are finally human enough.
We are not at the beginning of a new idea. We are at the moment a very old idea finally became real.
Five minutes well spent.