I want to write to you directly. Not as a consultant with a framework to sell, not as a futurist with a keynote deck, but someone who has spent the better part of four decades sitting across from senior leaders in banks, law firms, governments, and consulting practices, watching the gap between what executives say in public and what they genuinely fear in private.
Right now, that gap is enormous. And as AI adoption accelerates, it’s doing real damage..
What I Keep Hearing in the Room
In the last 18 months, I’ve been running leadership programs and communication workshops with executive teams navigating AI adoption across financial services, SaaS, data centres, and the public sector. Something has shifted in those rooms. The anxiety is no longer background noise. It’s present. I have seen this in the faces and bodies of those sitting in those chairs.
The analysts wonder if their role will exist in five years. The senior associates wonder if the leverage model that was supposed to reward them for a decade of hard work is going to evaporate before they get there. The policy advisers and compliance managers are watching their organisations quietly pilot tools that do in minutes what they have spent careers learning to do in weeks.
And the executives? Most of them are saying almost nothing.
I understand the instinct. I genuinely do. When I’m working with a leadership team that is itself uncertain, when the board is asking hard questions, when the strategy is still being formed, the temptation to stay quiet feels like the smart thing to do. Don’t alarm people. Don’t make promises we can’t keep. Wait until we know more. Not yet. This is becoming the rationale.
But here is what I’ve learned from working with leaders in moments of genuine disruption: silence is never neutral. Your people are not waiting in a comfortable holding pattern. They are filling your silence with the worst version of events they can imagine. And the most capable among them, the ones you most need to keep, are already updating their CVs.
Something Fundamentally Different Has Happened
I want to be careful here, because I’ve watched the tech hype cycle rise and fall before. The breathless predictions that never quite materialise. The technology that was going to transform everything mostly became a better search function.
This time is different. And the difference arrived in the last six months in a way that even seasoned sceptics are struggling to dismiss.
Until recently, artificial intelligence was essentially a very sophisticated autocomplete. Impressive, genuinely useful, but reactive. You asked it something, it answered, and it forgot you existed. What has happened since mid-2024 is something categorically different. We have moved into the era of agentic AI, systems that don’t just respond to questions but plan, act, and execute across multiple steps without human intervention at every turn.
Think about what that means in practice for knowledge work, for automation, and for how organisations increasingly automate tasks that once required teams of people. It’s no longer a tool that drafts a report when you ask it to. It’s a system that can be pointed at a goal: review this contract portfolio for clause risk, summarise the regulatory exposure, flag the top ten for senior review, and log the rest. And it will do that. Autonomously. Working through your systems, accessing your data, completing the loop. While you were in a meeting.
Microsoft, Google, OpenAI, and Anthropic all shipped major agentic capabilities in 2024 and 2025. UiPath has the first Business Orchestration platform that is ecosystem-neutral. Frameworks for multi-agent orchestration, where fleets of specialised AI agents hand tasks to each other like a well-run back office, went from experimental to production-ready. Gartner is predicting that by 2028, 33% of enterprise software applications will include agentic AI. Right now it’s less than 1%.
That is not an incremental change. That is a step-change in what white-collar work looks like and in the impact of AI on professional roles.
I was in a briefing with a major Australian super fund’s transformation team late last year. They were piloting an agent that could handle a compliance review workflow end to end, the kind of work that had previously occupied a team of eight across three weeks. They weren’t sure what to do with that information yet. But they knew they couldn’t unknow it. Eight people, effectively replaced by one agent, doing in an hour what once took weeks, with almost zero errors.
That conversation is happening in every institution I work with right now. The question isn’t whether this will affect us. The question is how fast and what we tell our people.
The Honest Assessment: What We Know. What We Don’t
I think executives owe their teams honesty about both sides of that ledger. So let me try to model it.
What we know with reasonable confidence:
AI in the workplace is already compressing tasks, reshaping roles, and changing expectations faster than many organisations are prepared for. Document review, first-draft generation, research synthesis, financial modelling, regulatory scanning, precedent search; these are not disappearing, but the hours they justify are shrinking. Work that took three weeks of associate time is heading toward hours or a few days. This is not a forecast. It’s already happening in the firms I work with.
The shape of roles is changing faster than the number of roles, for now. Most organisations are not yet running AI-driven redundancy programs. They are changing what they expect from people: less production, more judgement, more direction, and more accountability for outputs generated through AI assistants and other tools. That is a significant skills shift, and it requires support for both the human workforce and the leaders guiding it.
Entry-to-mid-level knowledge work faces structural pressure over a three-to-seven-year horizon. The leverage model, where law firms, consulting practices, and banks build profitability on a pyramid of junior staff doing high-volume analytical work, is under genuine strain. That is not a scare tactic. It is an arithmetic problem.
What we genuinely don’t know:
The medium-term trajectory. No one knows.
The models are improving at a pace that makes five-year workforce forecasts essentially fictional. Anyone who tells you with precision what a law firm or a government department looks like in 2030 is extrapolating from a trend that has changed shape four times in the last 24 months. I’ve learned to be suspicious of confident predictions in either direction.
We also don’t know if agentic AI deployment will go smoothly.
Gartner is also predicting that over 40% of agentic AI projects will be cancelled by end of 2027. Not because the technology doesn’t work, but because most organisations aren’t ready for it. The integration is hard. The governance is hard. The change management, which is really what I spend my working life on, is very hard.
Whether productivity gains translate to role preservation or role reduction. That is not a technology decision. It is a leadership decision. And it’s one that many executives I know are quietly deferring.
What Your People Need to Hear About AI in the Workplace
What employees want from leaders in a period of rapid AI in the workplace adoption is not certainty, but honesty about what this means for their work and their employee experience.
They’re not asking for a strategy document. They’re not asking you to predict the future. In my experience, what people in the middle of uncertainty actually need is threefold.
Acknowledge What People Already See
Firstly, they want to know you see what they see.
They want to know that leadership is not operating in a parallel reality where everything is fine, and AI is just a shiny new tool rather than a force reshaping the future of work. Acknowledgement, real acknowledgement and not performative, is the first thing that reduces anxiety.
Be Clear About What You’re Doing
Second, employees want to know what you’re doing about it.
Are you piloting tools? Which ones? What are you learning? Are you investing in reskilling or just watching? Specificity signals engagement. Vagueness signals avoidance.
Be Honest About What Happens If It Gets Harder
Finally, they want to know how you’ll handle it if it gets harder.
They’re not asking you to guarantee their jobs. They know you can’t. What they’re asking is for a commitment: that you won’t let restructuring happen in the dark, that you’ll be straight with them, that they’ll have enough warning to navigate their own futures intelligently.
That last one is where most executive communication falls short. Because it requires making a promise that has real accountability attached to it. And that’s uncomfortable.
Where I See Executives Getting This Wrong
Again, three patterns come up again and again in my work.
Mistake 1: Over-selling the Opportunity
The first is the enthusiasm offset. Leadership communicates AI in terms of possibility, innovation, and competitive advantage: “This is going to make us faster, better, more impactful”, without ever acknowledging the accompanying anxiety. The message lands as tone-deaf at best, dishonest at worst. Your people are not stupid. They can hear what isn’t being said. They have YouTube. TikTok. They have ChatGPT, Claude or Co-Pilot. People know what’s going on.
Mistake 2: Treating a Town Hall as Communication
The second is the one-and-done town hall. A CEO presentation, some slides about transformation strategy, or a Q&A with three safe questions. This is not communication. It is the performance of communication. The anxiety it produces in the weeks after, when nothing changes, and nothing is heard again, is often worse than if nothing had been said at all.
Mistake 3: Using Reskilling to Avoid Honesty
The third is the reskilling program as a substitute for honesty about AI adoption. I’ve seen organisations launch elaborate AI fluency training as a way of signalling commitment to their people, while simultaneously running a confidential headcount planning exercise that assumes 20% reduction in certain roles over three years. The training is real. The intention may even be genuine. But it doesn’t resolve the underlying question, and when people eventually find out the full picture, the trust damage is severe.
What Good Actually Looks Like
It’s not a single communication. It’s a process. Sustained, honest, layered, two-way. It’s adult-to-adult respect.
It starts with naming the uncertainty plainly. “We are navigating something genuinely new. We don’t have all the answers. Here is what we know, here is what we don’t, and here is how we’ll keep you informed as that changes.” In my experience, this is one of the most underrated things a leader can say. It sounds like weakness from the inside. From the outside, from the team hearing it, it is integrity.
It requires being specific about what you’re actually doing, not what you aspire to do. Which tools are you piloting? In which parts of the business? What have you learned so far? What governance have you put in place? The more specific, the more credible.
It requires genuine investment in capability development. Not as a PR exercise, but because the shift from task execution to task direction, oversight, and judgment is genuinely difficult, and people need support to make it. I work with organisations on exactly this, and the gap between what people need and what they’re given is consistently wider than leaders realise.
And it requires creating real channels for the conversation to come back up. Executive communication about AI cannot be a broadcast. Your front-line teams have a sharper, more current read on what is actually changing than most leadership teams do. If you can’t hear them, you are making a strategy with incomplete information.
The Sector-Specific Reality
I won’t pretend the pressure is uniform. It isn’t.
- In banking and financial services, the agentic shift is hitting compliance, credit analysis, fraud detection, and retail operations first, showing the early impact of AI on regulated work. The digital workers have, in a very real sense, already arrived. The governance questions, who is accountable when an agent makes a wrong call, are ones regulators are starting to ask directly.
- In law, the leverage model is arithmetically vulnerable. The business case for large junior associate cohorts doing document-intensive work is being eroded from below. Senior partners, I speak with know this. Most haven’t told their associates.
- In consulting, clients are starting to ask sharper questions about what they’re paying for. Deloitte was embarrassed by the exposure of work done by an LLM, but charged at consulting rates in October 2025. The line between strategic insight and deliverable production is being challenged in ways it hasn’t been before.
- In financial advice, the middle market is being squeezed from two directions. AI-augmented robo-advice from below, regulatory complexity from above. The advisers who will survive are those who have a value proposition that goes beyond information provision. Most haven’t done that work yet.
- In government, the pace is slower, but the eventual scale is large. Policy analysis, procurement review, service delivery, and legislative drafting are candidates for agentic augmentation. The public sector’s traditional slowness is not protection; it is just a delay.
For each of these industries, the workforce anxiety is legitimate. The job of leadership is not to make it go away with reassurance that may not be warranted. It is to make it navigable with honesty.
A Final Word
I was born in the 1950s. I guess that makes me officially old. I’ve been doing this work long enough to have watched leaders navigate major disruptions. The arrival of computers, the first PC’s, GFC, digital transformation, COVID, and the shift to hybrid work. In every case, the leaders who came out with their credibility intact were not those who had the best predictions. They were those who maintained the most honest ongoing conversation with their people about what was actually happening.
This moment is no different, except that the pace is faster, the stakes for individuals are higher, and the future of work is arriving faster than many want to acknowledge.
Your people are watching what you do here. Not what you say in a town hall. What you actually do. Whether you speak plainly or manage optics. Whether you invest in them or quietly plan around them. Whether you treat them as people who can handle the truth or as a population to be managed.
That choice is yours. And it will define how your leadership during this period is remembered.
