OpenAI's AI Keypad and How to Actually Use It

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. No filler, no hype — just what actually happened in AI this week, put together by a man who now outsources his own opinions to a machine.

Right. There is a physical keyboard that has an AI key on it, and OpenAI just made it. And before you say "I don't need a keypad with an AI button, I am not twelve years old and this is not a toy" — give me ninety seconds, because there is something genuinely useful buried inside this thing that has nothing to do with the hardware itself.

The keypad is called the OpenAI Keypad. It's a compact input device, essentially a numpad with an AI key wired directly into ChatGPT. Press it, speak or type, and you're straight into a conversation — no browser tab, no app switch, no logging in. TechCrunch's reviewer tried it this week and concluded it's a lot of fun for coders and slightly mystifying to everyone else. Which is a fair summary. But here's the thing the review mostly skimmed past: the keypad works because it exploits a workflow pattern that most people are already doing wrong. And you can replicate that pattern right now without buying anything.

The pattern is this: most people use AI in a separate lane. They finish doing a thing, then they go to ChatGPT, type something, get an answer, close the tab, and go back to what they were doing. And because switching between contexts takes effort, they only do it when the friction feels worth it. Which means they're missing a hundred small moments where AI would actually be useful — because in the moment, it doesn't feel worth the faff.

What the keypad does is collapse that friction to nearly zero. The AI is just always there, one button away, no mode switch required. And the insight is: when you reduce the effort to almost nothing, your usage patterns change. You start using it for smaller, faster, more immediate things. Not "write me a whole report" but "what's the word I'm looking for" or "is this sentence clear" or "quickly, is my logic right here." Those micro-uses compound. They're what actually move the needle on whether AI helps you think better or just occasionally writes an email for you.

So. Here's how to get the same effect without a £150 peripheral.

Pick one tool you're already in all day. For most people that's a browser. For some it's VS Code or Notion or Slack. Go to that tool's settings and find out if it has a native AI shortcut or extension. For Chrome, you can install the ChatGPT extension and set it to open on a keyboard shortcut — something like Command Shift G, or whatever your hands remember naturally. Takes three minutes. Once it's there, muscle memory does the rest.

Here's a worked example of why this matters in practice. Say you're writing a client proposal. You get to a paragraph where you're not sure whether you've explained something clearly or whether it sounds like jargon soup. Normally, you'd push on, file the uncertainty somewhere in the back of your head, and maybe fix it later — or more likely, not fix it. With the shortcut habit, you highlight the paragraph, hit your shortcut, paste it in, and type: "Is this clear to someone who doesn't work in this industry? If not, what's the confusing bit?" Twelve seconds. You have an answer. You fix it on the spot. The proposal is better. Nobody ever knows there was a wobble.

That's it. That's the use case. It's not that the AI is doing the work — it's that the AI is available exactly when you need a second opinion, and the habit of reaching for it is in your fingers rather than in a browser tab buried under seventeen other things.

Where does this break down? A few places worth knowing. First: if you don't already have a clear sense of what you're asking, easy access doesn't help. You'll just ask vague questions and get vague answers faster than before. The shortcut habit rewards people who have a half-formed thought and need it tested — not people who don't know what they want yet. Second: coders are going to get much more out of this than generalists, because their questions tend to be specific, verifiable, and immediately useful. If your day involves a lot of ambiguous creative judgment, the returns are smaller. Third: if your organisation has data policies about what you can paste into external AI tools — and a surprising number do, quietly — make sure you know where the line is before you start pasting client materials into anything.

One idea worth sitting alongside this — there's a piece on harrysharman.com arguing that AI adoption succeeds at the speed of behaviour change, not at the speed of capability release. The argument is that the bottleneck has never really been what AI can do. It's whether using it becomes a habit before it becomes a chore. A keypad with one button is a very expensive way to solve a habit problem. A keyboard shortcut is free, and it's the same fix.

So. The one thing you can do tomorrow morning: go to your browser or your main working tool, set a keyboard shortcut to your AI of choice, and write it on a Post-it on your screen for the first week. Then use it for something small — not a big task, not a project, just a small moment of doubt or stuck-ness. Do that three times and the habit is basically formed. After that, you won't need the Post-it.

That's your lot for this Sunday. Small habit, surprisingly large returns — and no keypad required.

That's the news, filtered down to what actually matters this week. Back tomorrow, same arrangement.