You have sat through one. The iceberg slide, the burning platform, the caterpillar that becomes a butterfly. The speaker meant it. The room clapped. By Wednesday nothing had moved and the programme it was meant to launch is still stuck.
That talk lands worse in 2026 than it did in 2019, and not because your people got cynical. It is because they are tired in a specific way. They have absorbed reorganisations, tooling migrations, a hybrid-work renegotiation and now AI rollouts, frequently inside the same two-year stretch. They are not waiting to be inspired into changing. They have changed constantly and are worn out by it, and underneath that sits a quieter question about whether the expertise they spent fifteen years building still counts for anything.
Book generic optimism into that room and you lose the audience in the first four minutes. Here is what to look for instead.
Most "future of work" talks are 2019 material with new nouns
The commonest failure is not a bad speaker. It is a good speaker running a deck that was written for the digital transformation era with the vocabulary swapped. "Digital" became "AI," "disruption" became "agents," and the underlying argument, that change is constant and you should embrace it, never moved.
You can spot the swap from the outline.
| The 2019 deck with new nouns | Written for 2026 |
|---|---|
| "Change is the only constant" | "You have changed six times already and you are tired, and here is what that does to a team" |
| "Adapt or be left behind" | "Here is which parts of your judgement get more valuable, not less" |
| Case studies from Blockbuster and Kodak | A programme from the last eighteen months, including one that did not work |
| "AI will augment, not replace" as reassurance | The actual, uncomfortable specifics of what gets automated in this audience's job |
| Ends with a call to be courageous | Ends with one thing the audience can do on Monday without asking permission |
The tell is whether the talk can survive a hostile question. Ask a prospective speaker what they would say to someone in the audience who thinks their role disappears in eighteen months. The 2019 deck has no answer to that, because it was written for a world where the change was about tools rather than about people's standing.
Change fatigue and AI anxiety are different problems
They get bundled together and they need different handling, so it is worth knowing which one is actually in your room.
Change fatigue is depletion. People know how to change, they have done it repeatedly, and each round has cost them relationships, workarounds and time. The symptom is not resistance, it is compliance without energy: everyone attends the training, nobody adopts anything. A speaker who treats this as a motivation problem is prescribing exercise to someone with a broken leg.
AI anxiety is status uncertainty. It is not mainly about tools. It is about whether the thing you are known for inside the organisation still marks you out. That is why senior, high-performing people are often the most unsettled: they have the most invested in an expertise that is being repriced.
The speakers worth their fee name whichever one is present, out loud, in the first ten minutes. Naming it is what buys permission for everything after. A room that has heard its actual condition described accurately will listen to a path. A room that has been told to embrace the future will not.
Brief for one behaviour, not three themes
This is the part most organisers get wrong, and it costs more than the choice of speaker.
A brief that says "AI and the future of work, plus change resilience, plus a bit on culture" produces a competent talk about all three that changes none of them. A brief that names one behaviour produces something you can measure.
Write it in this shape:
- The behaviour. One sentence, observable. "Managers run a monthly conversation with each report about which parts of their work AI now touches." Not "managers feel more confident about AI."
- Who does it. Not "the organisation." The 400 people-managers in operations.
- What currently stops them. Usually not ignorance. Usually no time, no permission, or a fear of being asked a question they cannot answer.
- What you have already tried. A speaker who does not know what has failed will confidently recommend it from the stage.
- What is off-limits or live in the background. A restructure, a union negotiation, a recent redundancy round. A speaker who walks into that unbriefed will do damage you cannot undo.
Send that page to the speaker before you sign anything and see what comes back. A practitioner will argue with it. They will tell you the behaviour is too broad, or that the real blocker is somewhere else, or that a keynote is the wrong format and you want a workshop instead. That argument is the single most useful signal in the whole process. The four questions we use for intake are a longer version of the same page.
Telling a practitioner from a pundit
Both are articulate. Both have good slides. The difference shows up under a specific kind of question, because pundits describe patterns and practitioners describe costs.
Six things that separate them:
- Responsibility, not observation. Ask what they owned rather than what they studied. "I advised on" and "I ran" are different jobs, and both can be valuable, but only one of them has been on the hook when it went wrong.
- A subtraction list. Ask what they had to stop doing to make a programme work. Every real transformation kills something: a reporting line, a metric, a beloved process. Pundits only ever describe additions.
- Recency with specifics. Ask what they were working on twelve months ago and what surprised them. Currency matters more than usual right now; a 2023 answer is a different technological era.
- A failure they were close to. Not a famous corporate failure. Theirs. The willingness to describe one in front of your leadership team also tells you how they will handle the Q&A.
- Sector translation on the spot. Ask how the talk changes for a hospital system versus a bank versus a software company. Practitioners answer instantly because they have made the adjustment before. Pundits describe a principle.
- Numbers with the right amount of precision. Be suspicious of a clean "37% productivity gain" and reassured by "it took about two quarters longer than we told the board it would." Real programmes have ranges and slippage.
This is the same verification logic we apply across every category, and the ten kinds of AI speaker guide has the equivalent question for each one. Anyone can claim to be an AI expert. We check, and the checking is mostly this.
Match the speaker to the moment, not the topic
The right choice depends less on your theme than on what your organisation is actually living through.
- A reorganisation or new operating model. You need coherence. Someone who can frame the change as a logic rather than an announcement, ideally a practitioner who has built one.
- An AI rollout. Pair a speaker who demystifies the technology with one who handles how people absorb it. Pure hype is the worst outcome; pure doom is second worst.
- Return-to-office tension. This is a trust problem dressed as a logistics problem, and another presentation about collaboration benefits will make it worse.
- After redundancies. Tread carefully. The people still there are watchful and grieving. Relentless positivity reads as contempt. You want someone who can acknowledge loss without wallowing and then rebuild a sense of forward motion.
- A workforce absorbing change unevenly. Automation and restructuring do not land the same way across every group, which is worth designing for rather than discovering afterwards.
Whichever moment you are in, tell us what is actually happening inside the organisation rather than the sanitised version, and we will come back within 24 hours with verified speakers matched to it, fee and our 17.5% commission printed on the same line.
The bar is simple to state and hard to clear: your people should leave a little steadier than they arrived, and one specific thing should be different on Monday. If a speaker cannot tell you what that thing is before the event, they will not produce it on the day.