We see it in workshops.
Ask a room of managers and juniors: ‘Where do you go first when you’re unsure about something?’ There’s usually a pause. A few smiles. Someone says ‘ChatGPT.’ A couple of managers look slightly uncomfortable.
It’s rarely about big strategic decisions. It’s the everyday questions: ‘Is this good enough to send?’ ‘Can you make this sound more confident?’ ‘What does this client actually mean?’ ‘How would a senior approach this?’
Five years ago, those questions would have gone straight to a manager. Now they often go to a browser tab. Not because managers don’t matter – because AI is instant, and doesn’t require you to interrupt someone who already looks overloaded.
In agency environments – especially where deadlines move hourly and everyone’s juggling multiple accounts – speed matters. When you’ve got ten minutes before a deck goes out, you’re not looking for a mentoring moment. You’re looking for a quick sense check.
AI is very good at that first pass. It tidies wording, suggests a structure, throws out options, and gives just enough reassurance to move forward. So if part of a manager’s role has been being the go-to for quick answers, that part is shrinking.
What’s being exposed
We don’t see this as a crisis. We see it as exposure. In fast-paced teams, a lot of management time has been spent reacting. Reviewing. Tweaking. Rewording. Saying ‘yes, that’s fine’ or ‘change slide three’.
If that’s the main value someone experiences from you, a tool that responds in seconds will naturally look efficient.
Here’s how it plays out in the room. A rising talent says ‘I ran it through AI, and it said it was strong.’ The manager reads it and replies: ‘It’s fine, but it doesn’t actually answer the brief.’ You can feel the gap. The tool offered confidence. The manager sees context. Context is the job.
AI doesn’t know your client history. It doesn’t know that this stakeholder hates risk. It doesn’t remember the political tension from last quarter. It doesn’t pick up on tone shifts in a call. That still requires judgement.
What better teams are doing
The strongest teams we’re seeing aren’t banning AI, and they aren’t pretending it solves everything. They’re getting clearer about standards.
One account director we worked with recently started documenting what ‘ready to send’ actually means for client comms. Not vague principles – specific checks. Does it answer the brief? Is there a clear recommendation? Are risks acknowledged? Once that was written down, the endless reassurance loops reduced. People still used AI to draft, but they measured work against a defined bar.
Another pattern: one-to-ones are shifting. Less ‘what did you deliver?’ and more ‘how did you decide?’. Managers asking: ‘What options did you rule out?’ ‘What made you choose that route?’ ‘What pushback are you anticipating?’ That builds judgement. A tool can suggest an answer. It can’t help someone think better over time.
There’s also an honesty piece. In some teams, AI use happens in the background and no one names it. People paste in polished copy and hope no one probes too hard. That creates tension. In healthier teams, someone will say ‘I used AI to get started, but I’m not convinced by the angle.’ The work still gets challenged properly. Standards stay intact. Transparency matters more than a policy document.
The question underneath
If AI is becoming the first port of call for direction, it’s worth asking why. Do people genuinely know what good looks like here, or are they guessing based on past reactions? Are managers available for thinking conversations, or mainly firefighting? Is feedback regular and specific, or mostly saved for formal reviews?
AI is brilliant at filling gaps. If expectations are fuzzy, it’ll produce something that sounds plausible. If managers are stretched, it’ll provide instant guidance. But it won’t build confidence. It won’t develop judgement. It won’t shape someone’s professional instincts.
If a manager’s main contribution is approving work and giving surface-level tweaks, that role will feel thinner over time. If their contribution is building clarity, confidence, and decision-making capability, it becomes more valuable.
The questions people are typing into AI aren’t dramatic. Small, slightly vulnerable ones: ‘Is this good enough?’ ‘Am I overthinking this?’ ‘What would someone more senior do?’ ‘Is this worth pushing back on?’ Those questions used to strengthen the manager relationship. Now they’re often handled elsewhere.
Competing on speed isn’t the answer. Being better at context, standards, and judgement probably is.
Practical steps teams can take
Write down what good looks like.
Clarity reduces unnecessary dependence on external answers.
Make one-to-ones about reasoning, not just updates.
Ask how someone approached the work.
Create simple guidance on AI use.
Not a complex policy – just clear expectations.
Normalise transparency.
It should be safe to say how AI was used and why.
Invest in manager development.
Coaching skills and confident communication matter more than ever.
AI will keep evolving. The question is whether our leadership and culture evolve with it.
At HappyHQ, we help agencies and in-house teams build strong cultures and the leaders and teams who sustain them. Find out more about how we work at happyhq.co.uk/what-we-do, or say hello at hello@happyhq.co.uk