software / human-computer interaction / AI / design / personal history

The First Function

Before assistants could act on our behalf, there was a little key that promised to help us find our way.

I always told folks: no matter which software you operate, always start with F1.

There’s a reason that’s the first function.

Get help.

We are all in this together.

I got my first computer in 1999. AltaVista. mIRC. Yahoo Mail. Floppy disks. The internet existed, obviously, but being online wasn’t yet the permanent condition of owning a computer. Sometimes connecting was an event in itself.

When I sat in front of unfamiliar software, wondering what the hell to do next, there was a key I could press.

F1.

The software opened its own help. No leaving the application to search the web. No watching twelve minutes of video for thirty seconds of instruction. Just the people who built the thing attempting to explain how to use it.

Sometimes the documentation was dreadful. But the principle was beautiful: the people building the software accepted some responsibility for helping you understand it.

When help became a destination

Over the next quarter-century we invented more and more ways to ask for help. Search engines, forums, Stack Overflow, tutorial videos, chatbots. All useful. All remarkable.

But look at the journey. You get stuck inside an application. You leave it, open a browser, describe the problem, find an answer for the wrong version, return, and try again.

We built entire industries around explaining the things we’d already built.

Somewhere along the way, help became a destination.

I think it should be a function.

The floppy disk that refused to die

I love that the floppy disk survived as a save icon long after the hardware disappeared from everyday life. People learned the symbol without ever touching the object.

It wasn’t merely nostalgia. A technical artifact had become a language for human intention.

There’s a distinction here I keep returning to: saving isn’t the same as storing.

Storage is what the machine does. Saving is what the human wants: I made something. Please don’t lose it.

A good interface translates between those worlds. The floppy icon did that remarkably well, even as applications moved toward autosave and other ways of preserving work.

F1 carried another human intention: I don’t know what to do next. Help me.

The new key

In January 2024 Microsoft announced the Copilot key, calling it the first significant change to the Windows PC keyboard in nearly thirty years.

Not because keyboards desperately needed another button. God knows we have enough.

But consider what the button represents.

Most keys tell the computer what to do: copy, paste, undo, save. F1 was the key you pressed when you didn’t know what to do.

Now the computer might not need us to know the command in the first place.

We can explain what we’re trying to accomplish. The software can respond in conversation. Increasingly, with our permission, it can act.

I don’t know if F1 will ever physically lose its place. Keyboards are museums of decisions nobody wants to revisit. But I do think the assistant is becoming the functional successor to the help key.

When help becomes action

Here’s where I get serious.

When I was younger I wanted to be a software developer partly because everyone around me wanted to be one. We didn’t really know what developing software meant.

I think about it differently now.

Someone might watch a film once or twice. Software runs through people’s lives every day: their work, their money, their relationships, their ability to find help. Most people will never meet the person who built it.

That is an enormous responsibility.

An old help page could give you the wrong instructions. An AI assistant might carry out the wrong instructions.

One error confuses you. The other might book the wrong taxi, send the wrong email, or commit you to something you never intended.

That boundary between information and action matters. We need systems that know when to explain, when to ask, when to act, and when to stop. We need to inspect their work, correct mistakes, and remain in charge of meaningful decisions.

An assistant isn’t valuable because it does everything without us. It’s valuable because it can do more with us.

The first function

My prediction is that assistants will inherit the role F1 once played: the first place we turn when we need help.

Maybe that help is an explanation. Maybe it’s a demonstration. Maybe it’s an action we explicitly authorize.

Whatever form it takes, I hope the builders remember the original promise. When someone doesn’t know what to do next, the software should help them find their way.

The first function was never really about a key.

It was about a responsibility.

Get help. Give help.

We are all in this together.


This is the personal companion to “Help Is a Function, Not a Destination” on the Teambotics blog, which explores what this idea means for building AI agents.