I’ve recently been talking to leaders about the conversations they’re having with their AI companions.  Do you have a voice version of your AI app? Mine is Copilot.

What I have come to realise is that with a private, always-available sounding board to help them think, test ideas and make decisions, a CEO or any other leader no longer needs to feel quite so lonely at the top. Which is useful, but it also means AI needs to be treated as a new source of influence, not just a convenient tool.

So, although your AI companion brings comfort, it also brings risk. It’s a team member, with access to your emails, calendar, channel chats and data, and the power to  influence you like a human team member.

And while improving, AI can’t be trusted to get its advice right. Yes, you can improve the chances you get good advice through good prompts, but it’s not guaranteed. Then there is the advice of others, often or exclusively guided by AI, that may be flawed.

AI is like email and smartphones. Both arrived with wondrous expectations, and before long we realised they had created new problems too, from email overwhelm to insidious doom scrolling.

That doesn’t mean you should avoid using AI. It means you need a few practical disciplines: know what data the tool can access, avoid putting sensitive material into uncontrolled environments, test important answers against other sources, and make clear to teams when AI can and cannot be used. Your challenge is not to slow AI down: It is to make sure people understand the boundaries before the boundaries are breached.

Enjoy your new companion. Use it to understand the risks that sit beneath the convenience, then put resources where they are most needed. Done well, AI can be a big plus for organisations, with far fewer minuses.