An AI keynote can leave an audience impressed and a business unchanged.
Before booking an AI speaker in Dubai, decide what your audience should be able to do differently afterwards. Recognise a valuable use case? Challenge an investment proposal? Understand where human judgement must remain in control?
Start with that outcome. Then choose the speaker.
I deliver AI keynotes, so I have a commercial interest in this subject. This is a proposed way to evaluate speakers, not an independent ranking. A good buying process should make it easier to choose someone else when their expertise better matches your brief.
Start with the audience and the decision
“We need a talk about AI” describes a subject, not a useful brief. Write one sentence instead: “After this session, this audience should be better able to…” Then make it specific enough to shape the content.
For a CEO retreat, the question might be whether to expand an AI initiative, redesign a process or keep a consequential decision under human control. A suitable outcome is that leaders can challenge the assumptions in an investment proposal: the intended value, available evidence, owner and exposure if it fails. A tour of the latest applications may contribute little to that discussion.
For an employee conference, people may need to understand what AI changes in their work and where to seek guidance. A useful outcome could be recognising an appropriate everyday use case and knowing when not to enter sensitive information into a tool. Avoid promising reassurance about jobs that the organisation itself cannot substantiate.
For a technical summit, participants may need to evaluate an architectural choice, understand a system’s limitations or compare evaluation methods. A useful outcome could be identifying the questions needed before testing an agent in an existing workflow. This audience may reasonably expect technical depth, including failure cases.
In Dubai, establish the audience’s working language, sector mix and familiarity with AI rather than assuming a single local perspective. Confirm whether the event brings together regional executives, an international audience or colleagues from one organisation. Those differences matter more than placing a skyline on the opening slide.
Decide which kind of presentation you need
An entertaining technology presentation can build curiosity and give a broad audience a memorable shared experience. That is a legitimate objective. It becomes a problem when entertainment is sold as evidence that a team is ready to make strategic decisions.
A technical briefing explains capabilities, constraints and implementation choices. It may be exactly right for specialists and wrong for a leadership audience whose immediate question concerns priorities or accountability.
A decision-focused keynote connects AI to choices: what to pursue, what evidence to request and what must remain someone’s responsibility. These approaches can overlap, but ask the speaker which one leads the proposed session. Energy and executive relevance are separate qualities; assess both.
Use a practical speaker-selection scorecard
Here is my proposed evaluation tool, not a scientifically validated system. Score each criterion 0: not demonstrated, 1: partly demonstrated, or 2: demonstrated with evidence relevant to your audience. Record the evidence and the unanswered question beside each score.
Apply these criteria to every speaker you shortlist, including me.
1. Business relevance
Can the speaker connect the subject to a decision your audience faces? Ask them to explain why one proposed topic belongs on your agenda. A strong answer names a business question and a useful takeaway; a weak answer simply repeats trends or fashionable terminology.
2. First-hand experience
Ask what the speaker personally did in the examples they intend to discuss. Distinguish implementation work, advisory work, teaching and reporting on someone else’s research. Each can be valuable. The test is whether the experience supports the claim, and whether the speaker can explain a limitation or an unsuccessful approach without hiding behind client logos.
3. Clarity of explanation
Use a recording to assess whether a difficult idea becomes understandable without becoming misleading. Could a non-technical executive explain the point to a colleague afterwards? For a specialist audience, clarity also means precision about what the system can and cannot do.
4. Audience-specific preparation
Look for questions about the audience’s priorities, existing initiatives, level of knowledge and sensitivities. Ask what would change between this session and the speaker’s standard presentation. A different title or a few references to your industry is not the same as changing the substance.
5. Usefulness after the event
Ask what people can use afterwards: a set of decision questions, a framework or an agreed follow-up discussion. Do not assume materials are included; clarify that before booking. Practical value does not require a large workbook, but it should be possible to describe what survives beyond the applause.
Use the scores to structure a discussion, not to manufacture certainty. Do not let a high total conceal missing expertise on a critical subject. Agree any non-negotiable requirement before comparing candidates, and mark unavailable evidence as “not yet demonstrated” rather than guessing.
Request evidence that reveals the fit
Four requests make a shortlist easier to assess:
- A relevant recording. Ask for a sustained segment with an explanation, example and, if available, audience questions. A polished showreel demonstrates stage presence, but edited applause cannot show how someone handles a difficult question.
- A proposed session outline. Look for a clear progression from the audience’s problem to the ideas and decisions the session will address. Ask what the speaker would leave out to protect that focus.
- Specific audience takeaways. “Understand the future” is hard to evaluate. “Distinguish an AI recommendation from a decision requiring human approval” gives you something concrete to discuss with the organiser and audience.
- A reference from a comparable engagement. With permission, ask an organiser about preparation, audience fit and responsiveness to the brief. Similarity of purpose matters more than the prestige of the organisation’s name.
Respect confidentiality. An anonymised example may be more appropriate than a named client, provided the speaker explains their contribution without turning unverified outcomes into a success story. If a reference or recording cannot be shared, discuss what alternative evidence would help.
Also confirm the practical terms: duration, Q&A, recording rights, travel, fees and cancellation arrangements. These are part of choosing a suitable engagement, even when the content is excellent.
Choose a keynote, working session or training programme
Recommend a keynote when the priority is a shared perspective across an audience: opening a conference, framing a leadership conversation or challenging assumptions. Agree the central message and the room’s starting point.
Choose a working session when a smaller group needs to examine a specific problem. Participants should bring context, and there must be time to question assumptions and compare options. If the task is to define an AI pilot or agree decision boundaries, a presentation alone may be insufficient.
Choose a training programme when people need repeatable capability: using approved tools, evaluating outputs or following a new process. That requires practice, feedback and a way to assess learning. A memorable keynote cannot substitute for those elements.
Combining formats can make sense, but only when each has a clear job. Avoid adding a workshop simply because it makes the package look more substantial.
A real adaptation: making room for executive participation
In an executive teaching experience I described publicly, I put a question to a room of directors. Nobody answered. The CFO was nearby. My reading of the silence was that answering exposed people to a reputational risk in front of colleagues.
My initial move was to ask a question openly, as I might in a classroom. I changed how I taught executives. I stopped correcting them publicly and presented ideas in a way that let them recognise an insight, rather than feel they had been caught missing something.
The reason was audience fit. A senior person’s willingness to participate can depend on how the discussion affects their standing in the room. That observation concerns the teaching method, not a measured business result.
For an organiser, the useful question is specific: how will this speaker make it possible for our audience to admit uncertainty? Preparation includes the dynamics of the room as well as the subject of the slides.
How I would prepare for an executive audience
My starting point would be the decision you want the event to support. I would ask about the audience, what it already understands, which AI conversations are live inside the organisation and what it should be better equipped to question afterwards. We would agree the outline and intended takeaways before settling on a title.
For a Dubai leadership event, I would propose three starting points:
- AI strategy: choosing priorities and testing assumptions before committing resources.
- Digital Employees: defining AI-enabled roles around objectives, permissions, escalation and human ownership.
- Human responsibility in AI delegation: deciding what a system may execute, what requires approval and who remains accountable.
These are starting points for a brief, not a claim that every audience needs the same talk. If your immediate need is technical implementation or sustained skills development, the selection process should make that visible too.
