AI & The Future of Work

Your new Slack coworker doesn't need a desk

AI just moved into your team's Slack. The question isn't whether your people are ready — it's whether your change plan is.

On 23 June, something quietly significant happened. Claude moved into Slack. Not as an app you open in another tab — as a teammate you tag. You write @Claude in a channel, hand it a job, and get on with your day while it works.

It's called Claude Tag, and it's in beta now for Team and Enterprise customers. Tag it with a request in plain English, and it breaks the work into steps, uses the tools and data you've connected, and reports back in the thread. It learns your context by following the channel, so you stop re-explaining yourself. It works asynchronously — set it going and it'll beaver away for hours, even days. The whole team shares one Claude, so everyone can see what it's doing and pick up where someone else left off.

I saw it land in my feed (via the always-excellent The Rundown) and my first thought wasn't "neat feature." It was: oh, this is going to be a people problem.

The third act

We've watched AI move through three acts now.

First, the chatbot — you went to it, in a separate window, and asked it things. Then the copilot — it sat beside you inside your tools, suggesting as you went. And now the coworker — in the room, in the channel, in the actual flow of your team's work.

That last step is a bigger leap than it looks. A chatbot is something you use. A coworker is something your team has to work alongside — and that changes everything about how people feel.

#project-rollout
PD
Priya
@Claude can you pull the go-live risks by team and draft a first-cut comms plan?
C
ClaudeApp
On it 👋 Breaking this into three steps — pulling the risk data, drafting the comms, then back in this thread. Give me a few minutes.
TR
Tom
Perfect — I'll drop the training dates in once Claude's draft is up.
You tag it, it picks up the work, the whole team watches it happen. That's a coworker — not a chatbot.

The bit every headline misses

Dropping an AI coworker into your team's Slack is not a tooling change. It's an organisational change. And like every organisational change in the last 20 years, it will be decided by the people side — not the technology.

Because the moment that bot appears in the channel, your people start asking themselves things they won't say out loud:

Leave those questions unanswered and you don't get adoption. You get quiet anxiety, awkward avoidance, and a very expensive tool that nobody quite trusts.

The technology moved in overnight. The trust doesn't. That gap is the entire job.

What a change plan for an AI coworker looks like

You don't need a 40-page strategy. You need to do the few things that actually move people from wary to willing:

  1. Name what it does — and what it doesn't. Clarity is the fastest way to kill fear. Spell out the tasks it's there for, and the ones that stay human.
  2. Give every output a human owner. The bot drafts; a person signs off. Accountability never moves to the AI, and everyone needs to know that.
  3. Set the norms. When do you tag it? What should never be shared in a channel it can see? Agree the etiquette before, not after.
  4. Protect psychological safety — it's now the law. Since December 2025, poorly managed change is a named psychosocial hazard under Australian WHS law. Watching an AI do tasks that used to be "yours" is exactly the kind of thing that needs support, not silence. (More on that here.)
  5. Build the capability. Show people how to work with it — how to brief it, check it, get the best from it. Skip this and they'll quietly route around it.

And if you don't have a plan?

Your people will make their own. Some will lean in and quietly outpace everyone. Others will avoid it entirely. A few will paste things into it they really shouldn't. That's not a hypothetical — it's the shadow AI story playing out in real time, just with a name and an avatar this time.

The only question left

Here's what's genuinely new: the technology is no longer the hard part. Claude Tag is live today. It runs on Claude Opus 4.8. It works. It'll be in other platforms before long.

So the question isn't whether the tool is ready. It's the question you've always faced with any big change — is your plan for the people ready? That's the work. That's what I do.

A good place to start is knowing where your teams actually sit on the readiness curve. The free Change Impact Scorecard maps it in about ten minutes.

Check your team's readiness — free Scorecard →

Spotted via The Rundown. Full details in Anthropic's announcement (23 June 2026).

Sheena Karim
Written by Sheena Karim Connect on LinkedIn ↗

Keep reading