AI and teams · 3 min read

The “proficient in Office” era of AI

I built our team a friendly bot so they wouldn’t have to touch Claude Code. A few weeks later I scrapped it and asked everyone to learn the real tool.

The bot

I built our team a friendly OpenClaw bot on Telegram, so they could use our AI workflows without touching Claude Code. You’d message it like a colleague, and it would run the same skills and automations underneath.

Why I scrapped it

I’d wrongly assumed that my team, mostly creatives, would struggle with a “code” tool. I quickly realised there’s no point insulating them from learning the new tools directly.

The difference between typing into my bot and typing into Claude Code turned out to be small. It exposes you to a bit more complexity, but at the end of the day you’re typing text into a prompt. What you get in return is everything: the shared skills, the connectors, the ability to fix a process yourself. Within a session or two, people on the team were asking how to update a shared skill themselves, which is exactly what the bot would have stopped them doing.

The ’90s CV line

It reminds me of the good old ’90s, when “proficient in Microsoft Word and Excel” was the kind of thing people put in the skills section of their CV. If you saw that today, you’d assume it was retro cosplay or performance art.

A skill stops being listed when it starts being assumed. Nobody lists email either.

In 2026, people flex their AI agent credentials in job applications. We’re in the “proficient in Office” era of AI.

What that means for hiring and training

If the skill is about to be assumed, the useful move is to teach the real tool early, to everyone, rather than build a softer version for the people you think won’t cope. Shielding people from it only delays the point where it’s simply part of the job.

How fast it fades

I expect this one will fade a lot faster than it did last time.

More: all writing · LinkedIn