A few weeks ago I chaired a Leadership Mind The Product evening in London. A room of senior product and engineering leaders sat down to talk about one question:
When recently were you pushing too hard, or not hard enough? What were the signs?
That's a deceptively simple question. In a room of executives it became something else entirely. It became a conversation about listening, about signals, and ultimately about how AI is changing what it means to lead well.
It was one of the most useful conversations I've been involved in about AI leadership all year. Everyone was asking such challenging and engaging questions.
Getting the room warmed up
Tempting as it was to get everyone doing star jumps, it was Tuesday evening after tasty food and some wine, so instead we started with a warm-up question: "In one sentence, what's a standard you hold that someone on your team probably finds annoying?"
Among many interesting angles, this answer was widely agreed with: "I won't accept a status update or a document unless it has a clear decision or an ask."
The room nodded, and shared they had their version of this standard.
The story that reframed the night
One leader told a story that, judging by the silence in the room, hit a lot of us in the chest.
A junior member of their team had been pushing for more responsibility and a step into management. The team needed the role filled. So they said yes. But deep down they knew this person wasn't ready.
What followed was months of hard work from the individual, and an emotionally draining experience for everyone. They never captured the respect of the team. They kept telling their leader they were failing. Their motivation got destroyed in the process.
"Now I don't even think they want to be in management at all. I did not set them up for success by doing the easy thing and giving them what they asked for."
The room went quiet. This is the version of pushing too hard that nobody warns you about. Not driving people too hard. Not setting unreasonable goals. Saying yes when your gut is telling you no, because the no is harder.
This story describes a recipe for burnout dressed up as opportunity. It happens because the leader didn't trust themselves and saw what initially appeared to be an easy path forward.
What the room kept coming back to
We went round, and as people spoke, four trends evolved.
The first was listening as a leadership skill, not a soft skill. Several people described moments where their team's discomfort was the data. We had stories of seeing the panic is someones eyes. The leaders who'd grown most weren't the ones who'd toughened up. They were the ones who'd slowed down enough to hear what was being said underneath what was being said
The second was managing the expectation that performance will drop during change. One leader put it bluntly, “when you're driving a transformation, the team's performance has to take a hit. And stakeholders need to be prepared for it. Pretending it doesn't is how you set people up to fail and call it underperformance.”
The third was the early signal. I asked the room: what's the earliest signal you trust now that you didn't trust before? The answers were not what I expected. They weren't metrics or velocity or roadmap slip. They were:
The final trend was the early signal. I asked for takeaways from our conversation. I was expecting something related to metrics, or velocity. Instead key takeaways included:
- Being on the look out of behaviour change such as a direct report going quiet in a meeting they'd normally drive.
- A peer paraphrasing your decision back to you wrong.
- The gap between what someone says in a 1:1 and what they say in a group.
- Your own urge to argue against a piece of feedback before you've understood it.
These are the telltales executives often dismiss as "soft”, but in a room of product leaders living AI transformation they were respected as the leading indicators.
Then we flipped to AI, and the conversation got serious
I asked, “how has AI changed the line between holding people to a high standard and being unkind?”
The feeling in the room moved. The energy lifted. It was refreshing to hear leadership speak open and honestly about how AI was making them or their teams feel:
- People are scared, many quietly. Mainly about job security. Typically about whether they'll be the one made redundant by the thing they're being asked to adopt
- Experts who have an identity in the craft was scared about their reputation, and confused about their future purpose. (eg communication experts, senior engineers)
- Leaders are scared too. Fewer admit it, but you could feel it. A few times the phase “Adapt or die” was used, but this is not a strategy. It's a fear stated as a slogan.
- Compliance and risk teams are pulling in the opposite direction to the build teams. Someone described any AI tool with a free-text field triggering a compliance alarm, and the resulting standoff being treated as a process problem when it was actually a trust problem.
The last person to speak summed things up perfectly: "We have to level up our human game to lead past the fear."
That's the work. AI has not changed what good leadership is. It has raised the price of bad leadership. Later over networking drinks an executive shared a view “AI is removing the hiding places for bad decisions. Leadership has to improve”.
My thoughts as leadership in the AI era strengthens a communication problem.
Here's what I think the room was circling without quite naming.
Most leadership advice is still pitched at the wrong layer. It tells you how to set strategy, how to write vision, how to demand accountability. All useful. None of it is the bottleneck right now. The bottleneck is communication and alignment, both are the core of leadership effectiveness.
When AI is reshaping how your team works, what their roles mean, and how individuals feel worth, the cost of missing sub optimal communication is higher than it has ever been. The leader who said yes to a promotion they knew wasn't right? That's the pattern, scaled up across every team adopting AI right now. Saying yes to the easy path. Skipping the preparation to land communication with your audience, repeating the AI response, and not listening to the humans infront of you. Alignment needs emotional connection, and this, for now, is a human thing not and AI thing.
I am lucky to see how many different leaders operate across multiple organisations. The following shares three patterns I see from exceptional leaders driving AI transformation.
They watch the human signals before the metrics move. By the time engagement scores drop, you are already six weeks late. The early signal is in the meeting, in the silence, in the quality of the questions people are willing to ask. If you are only reading dashboards, you are leading on lagging indicators.
They are honest about the true cost of transformation. They name the cost out loud, to performance, to confidence, to identity, and protecting people through it. Positivity is not a strategy. Honesty about cost is.
They lead the fear, they do not manage around it. Fear that is named loses about half its power. Fear that is suppressed compounds. The leaders making AI work in their organisations are not the calmest in the room because they are unbothered. They are the calmest in the room because they have done their own thinking about what they are afraid of, and they can hold space for other people's version of the same thing.
The takeaway I hope you leave with
We talk a lot about the technology. There is daily news about some new model, or feature in Claude, or the latest OpenClaw security disaster. The conversation we are not yet having loud enough about AI in our organisations is not a technology conversation. It is a leadership conversation, and specifically a listening conversation.
The leaders who win the next two years will not be the ones who adopted the tools fastest. They will be the ones who read the signals best while the tools were being adopted around them, and adapted their behaviour to what it told them. Adapt to win.
If you are a senior product or engineering leader and you have read this far, here is the question I would leave you with. Where are you currently saying yes when your gut is saying no, and what would it cost you to trust the signal instead?
That's where the work is.
