You booked an "AI futurist" in 2023. They showed a demo, made three predictions, and closed on how nothing would ever be the same. Four hundred people nodded. Nobody did anything differently on Monday.
That talk does not work in 2026, and the reason is simple: your audience has used the tools. They have written prompts, been disappointed, found a workaround, and quietly built shadow processes your IT team has not approved. They are past wonder. They want to know what to do, what not to bother with, and who is telling them the truth.
Most speaker rosters have not caught up. Here is how to read the market.
The tell: is this a 2022 speaker with the nouns swapped?
There is a whole cohort of speakers who were "digital transformation" experts in 2018, "blockchain" adjacent in 2021, and "AI" now. The deck structure is identical. Only the nouns changed.
Four signals give it away:
The bio leads with attention, not with work. "Thought leader," "influencer," a follower count, a TEDx from years ago, a magazine list. None of that is evidence. It is the easiest part of a bio to acquire and the least correlated with whether the talk is any good.
Nothing has a date on it. A credible 2026 speaker can point to something specific from the last twelve to eighteen months: a system they shipped, a paper they published, a rollout they led, a company they are running right now. If the most recent concrete thing is 2022, you are booking a historian.
The talk is about "AI," full stop. Generalist AI talks have almost no remaining value, because the general story is now common knowledge. The value has moved into specificity: your sector, your regulator, your failure modes, your procurement cycle.
There is no failure in the story. Anyone who has actually deployed this technology has lost a quarter to a bad rollout, watched a pilot stall in legal, or rebuilt a workflow twice. A speaker with only wins has either not done the work or will not be candid on your stage. Ask for a failure early. It is the fastest disqualifier there is.
Anyone can claim to be an AI expert. The whole job, ours and yours, is checking. We verify every speaker individually before they are bookable, which is how we work, but the checks below are ones you can run yourself in an afternoon.
The five archetypes, and what each one is for
Almost every AI speaker is one of five things. Knowing which one you are buying prevents the most common booking mistake, which is hiring a visionary when the room needed an operator.
| Archetype | Book them when | Do not book them when | The verification that matters |
|---|---|---|---|
| Practitioner-founder | You need credibility with skeptics, engineers or anyone who has been pitched too often | You need a broad, energizing open for a mixed audience | A product or deployment you can look at, and a straight answer about what failed |
| Researcher | The room needs to understand what the technology can and cannot actually do, and why | You need a Monday-morning playbook | Published work with dates, and evidence they can talk to non-specialists |
| Sector specialist | Your audience shares an industry, a regulator or a failure mode | Your audience is cross-industry and the specifics will not land | They can name your sector's constraints unprompted, before you brief them |
| Workplace psychologist | Adoption has stalled, people are anxious, and the bottleneck is human | The room wants technical depth | Practice with real organizations, not just research on organizations |
| Policy and governance voice | A board, a regulator or a risk committee is watching | You need energy in a sales kickoff | Current on the regulation as it stands this quarter, not last year |
Practitioner-founders
The builders. They have shipped something, run something, or are mid-build right now. The material is current by construction, because they were doing it last week.
Verify: ask for the thing. A product you can open, a repository, a customer, a deployment inside a company you have heard of. Then ask what broke. A founder who says "the model was easy, the data access took nine months" is telling you the truth about enterprise AI. One who describes a clean rollout is describing a case study, not a project.
Watch for: the founder whose company is the talk. There is a difference between someone who draws on their work and someone delivering a 40-minute pitch you paid for. Ask directly whether they will name their product on stage, and get the answer in the contract.
Researchers
The people who know what the technology actually does, as opposed to what the marketing says. Enormously valuable for cutting through nonsense, especially with senior audiences who have been sold to hard.
Verify: the published record, with dates. Then a much harder check — watch them explain something technical to a non-technical audience for at least ten unedited minutes. The single most common failure with researchers is a brilliant person who cannot get out of their own vocabulary. Plenty can. Verify which kind you have before you put them in front of 400 salespeople.
Sector specialists
The band most under-supplied and most under-booked. A talk about "AI in healthcare" beats a talk about "AI" for a room of hospital administrators, every time, and it is not close.
Verify: before you brief them, ask them what they think your sector's biggest constraint is. A real specialist will name procurement cycles, a specific regulation, a data-residency problem or a union agreement without prompting. A generalist will say "change management." That one question does most of the work.
Watch for: the specialist whose sector knowledge is five years old. Regulation in particular moves fast enough that a 2024 view is actively misleading. Ask what changed in their sector this year.
Workplace psychologists and adoption specialists
Your real problem is often not the model. It is the 300 people who are anxious about their jobs, skeptical of the last three initiatives, or already using tools nobody approved. Technology talks do not touch any of that. These speakers do.
Verify: ask whether they have worked with organizations, not just studied them. Ask for a specific case where adoption failed and what they would do differently. And check the tone: a room that is genuinely worried about redundancy can smell reassurance that has not been earned from the back row. The workplace transformation guide goes deeper on choosing this type.
Policy and governance voices
Increasingly non-optional for boards, public-sector events, regulated industries and anyone whose risk committee has started asking questions.
Verify: currency, above all. Ask them to describe the regulatory position in your jurisdiction as of this quarter and what they expect to change next. Ask whether they have advised an organization through an actual compliance process or only written about it. Both are legitimate, but they produce very different talks, and only one of them gives your risk team language they can reuse.
Matching the archetype to what you need to happen
Start from the outcome, not the topic. The topic is always "AI." The outcome is what you are actually buying.
- Executives are stalling. A practitioner-founder or sector specialist who can talk peer-to-peer with senior leaders and make the cost of doing nothing concrete.
- Adoption has stalled mid-rollout. A workplace psychologist, or a practitioner who has personally managed the human side of a change.
- The room is full of skeptics. A practitioner with receipts, including the failures. Nothing else survives an engineering audience.
- The board or the regulator is watching. A governance voice, paired with someone operational so the session is not purely defensive.
- The event has felt repetitive for three years. A voice from outside the usual circuit. Not as a gesture. Because the use cases are genuinely different, which is the argument in our post on diversity in AI speaking.
If you would rather not sort this yourself, that is the job we do. Send us four answers about your event and you get three to five verified speakers back within 24 hours, with the archetype named and the reasoning shown. No account, no sales call.
A useful gut check before you sign anything: write the one sentence you want an attendee to repeat to a colleague the next day. If your chosen speaker cannot plausibly produce that sentence, you have the right person for someone else's event.
Then run the diligence. Unedited footage from the last year, one dated artifact of real work, two references you actually call, and a straight answer about a failure. That is four checks, and it takes an afternoon. The rest of the process — the brief, the quote, the contract, the week-of logistics — is in how to book an AI speaker, and what each archetype typically costs is in the 2026 fee guide.