We're rolling out AI" can mean three different futures for your people.
Published Date:
June 9, 2026
The exec meeting was forty minutes in before anyone noticed the problem.
A founder I was working with - PE-backed, around 200 people - had "AI rollout" on the agenda. Everyone around the table was nodding. So I asked each of them, in turn, what they understood the rollout to be.
The CFO described a tool: something the finance team would prompt to speed up reporting. The COO described an agent: software that would handle a chunk of customer queries end to end, with limited human involvement. The CEO described something closer to a philosophy.
Three leaders. One word. Three different futures for the people who worked there - and three very different conversations that would eventually need having with them.
I thought about that meeting this week when I noticed how much effort law firms are now putting into glossaries for their business clients - patient explanations of the difference between an AI model, a platform, an agent, an application, a system and a tool. The lawyers' reason for caring is sound: the distinctions carry real implications for risk, liability and how emerging regulation will apply.
All true. But there's a reason to care about the vocabulary that sits well before the lawyers get involved - and I'd argue it's the bigger one.
The distinctions aren't pedantry. They map, almost one to one, onto how much of someone's working life is about to move.
A tool - the spell-checker kind, the prompt-it-and-get-an-output kind - changes how a task gets done. The person is still doing the job. They may even enjoy it more.
An agent is different. An agent plans, decides and acts - completing multi-step workflows with limited human input. That isn't a faster way of doing the job. That's a redrawing of where the job's edges are.
And a system - multiple models, tools and data stitched together to automate an entire process - is a workforce question wearing a technology costume.
I spent five and a half years as Global People Director inside a fast-growing AI consultancy, and one of the quieter lessons of that time was this: the technical people almost always knew which of these they were building. It was the non-technical teams and sometimes senior leaders that flattened everything into "AI" - and that's where communication issues can arise.
Here's the uncomfortable bit. When a leadership team says "we're rolling out AI" without saying which kind, the vagueness isn't neutral. It's load-bearing.
It lets you announce ambition to the board without committing to a people plan. It lets you defer the question of whose role changes, because nobody can pin you to specifics. It feels safer.
But your people don't experience vagueness as safety. They experience it as a blank space - and they fill blank spaces with the worst plausible version. The customer service team that hears "AI agents" in a town hall and nothing else will not assume you mean a pilot in one queue. They'll assume you mean them, soon, all of it.
So the gap opens up: between the word in the board deck and what actually lands on someone's Tuesday morning. And every week that gap stays open, it costs you something you'll want later - the credibility to say "this part of the change is genuinely good for you" and be believed.
There's a harder edge to this too. If the honest answer is that an agent or a system will eventually mean fewer roles, the vague version doesn't make that easier. It makes it worse - because by the time you get to formal consultation, your people have already concluded the vagueness was cover. In ER terms, you've spent the trust before the process even starts.
So the practical turn is less about technology governance and more about leadership discipline.
Before the next all-hands, the next investor update that mentions AI, the next line in the strategy deck - get the leadership team to answer, in plain words: is this a tool, an agent, or a system? Which tasks does it touch, which roles does it touch, and what's true about that today versus what's speculative?
Then say the specific version out loud to your people. "We're piloting a tool that drafts first-pass responses in the support queue; humans still own every reply; here's what we'll review in three months" is a sentence a workforce can work with. "We're embracing AI" is not.
Notice that the precise version also forces you to confront what you don't yet know. If you can't say whether the thing is a tool or an agent, you don't have a communication problem - you have a decision you haven't made. Better to find that out in the leadership meeting than in the consultation room.
And if the precise answer carries hard news in it - some roles will shrink, some will go - then precision is still the kinder route. People can plan around a specific future. Nobody can plan around a fog.
The lawyers are right that the definitions matter for liability. But in the Human + AI era, the first audience for your AI vocabulary isn't a regulator. It's the person three layers down wondering whether they should be worried - and reading your word choice for clues.
Being specific won't make every conversation comfortable. It will make every conversation possible.
If you're somewhere in the middle of this - an announcement made, a fog forming - and want a sounding board before the next town hall, a 30-minute conversation is free and carries no obligation.
A note on the consultation points above: this is a practitioner's view, not legal advice. If your AI plans may lead to redundancies or significant changes to roles, take advice from an employment lawyer on your consultation obligations before you act.
Fractional HR, executive coaching and workforce transformation — designed for the Human + AI era.
