You came here for ten names. You are going to get ten kinds of speaker instead, and I would rather explain that swap in the first paragraph than bury it under a list.
We are new. iShruti launched in June 2026, we verify every speaker individually before we will put them in front of you, and we are not going to publish a named top ten built partly out of people we have not checked. Legacy bureaus publish those lists constantly. Read a few closely and you will notice most of them are some blend of who is under contract and who has the biggest following. That is a popularity contest with a headline on it. It is not a match for your room.
The category is the durable part. Names churn every quarter; the ten jobs below have been stable for two years and will hold through 2027. Once you know which job your event needs done, the name is the easy half of the problem.
So here is how to use this. For each kind: what it is genuinely good for, what it is wrong for, and one question that tests whether the person in front of you actually is that kind. The questions matter more than the categories. Anyone can claim to be an AI expert, and a fluent generalist can present as any of these ten for about four minutes. We check. You should too.
If you want the shorter sort, the five AI speaker archetypes collapse this into a five-minute read. This one goes wider because Fortune 500 programmes usually need more than one of these across a year.
Start from your problem, not from the topic
The most expensive booking mistake is hiring a visionary when you needed an operator. Both are excellent speakers. Only one of them fixes what is actually wrong.
| If the real problem is | Book | Do not book |
|---|---|---|
| The board thinks AI is one undifferentiated thing | Frontier translator | Practitioner-founder |
| People are quietly frightened about their jobs | Workplace psychologist | Futurist of any kind |
| "That won't work in our industry" | Sector specialist | Generalist educator |
| Policy is running behind the programme | Governance voice | Frontier translator |
| Global workforce, US-shaped assumptions | Global South practitioner | Domestic future-of-work strategist |
| Nobody uses the tools you already bought | Educator | Futurist |
| The programme has stalled 18 months in | Transformation lead | Practitioner-founder |
1. The frontier translator, for when the board's mental model is two years stale
Good for: strategy offsites, board sessions, opening keynotes. The job is calibration. What is real inside an eighteen-month window, what is a demo with good lighting, what is a five-year bet you should fund but not plan around.
Wrong for: a room that has already committed and needs to execute. Also wrong in front of a technical audience, who have read the same papers and will spend the Q&A correcting slides.
Test question: "Name a capability you were confident about eighteen months ago that turned out to be further away than you thought." Someone who has never revised a forecast in public is selling certainty, and certainty is the one thing nobody can currently supply.
2. The workplace psychologist, for the fear nobody says out loud
Good for: all-hands after a rollout announcement, HR and people conferences, any room carrying two years of accumulated change fatigue. They explain the mechanics of how people absorb change and what rebuilds trust once it has gone.
Wrong for: technical roadmap sessions. Also wrong as your only move if leadership has not answered the substantive questions about roles and job security. A talk about feelings cannot paper over a policy vacuum, and an audience will spot the attempt inside five minutes.
Test question: "What is the most common mistake leaders make in the first week after announcing an AI programme?" You want a specific observed behaviour, not "communicate more."
3. The sector specialist, healthcare being the hardest version
Good for: single-industry audiences who have heard the generic AI talk twice and have stopped listening. In healthcare that means someone fluent in clinical workflow, regulatory clearance and the liability picture, not someone who uses hospitals as a tidy example.
Wrong for: mixed-industry conferences, where depth turns into noise for two thirds of the room. Also wrong if your sector is genuinely early, because a specialist with no deployments to describe is a generalist with better vocabulary.
Test question: "Walk me through one deployment in my sector that got killed, and why." Describing a failure up close requires having been near one.
4. The governance voice, for when the programme is ahead of the policy
Good for: legal and compliance leadership, board governance sessions, risk committees. The EU AI Act is in force, enterprise procurement questionnaires already ask about model provenance, and most organisations are shipping faster than their own rules.
Wrong for: a motivational slot, and wrong if what you actually want is someone to bless what you are already doing. That is not a keynote, that is an audit you are hoping to fail upward through.
Test question: "Where do you think the current rules are wrong?" Someone who can only recite regulation is a summariser. You can read the summary yourself.
5. The Global South practitioner, for the markets that are not on your slides
Good for: multinationals, global supply chains, and any organisation whose growth markets are not the ones its strategy deck was written about. Deployment under low bandwidth, multiple languages, thin data and hard budget ceilings teaches lessons that travel straight back to enterprise teams facing the same constraints at a different scale.
Wrong for: being booked to represent a region. Brief them on their work, not their passport. We wrote about why the geography of your lineup changes its accuracy rather than just its optics.
Test question: "What does your work assume about infrastructure that a US audience would get wrong?" A good answer will be uncomfortable and specific.
6. The financial services specialist, for the most regulated deployment on earth
Good for: banks, insurers, asset managers, and the risk, fraud and compliance functions inside them. Model governance, explainability requirements and supervisory expectations are their native language, and that is a genuinely different conversation from clinical AI or industrial AI.
Wrong for: general business audiences, who will drown in the second half. Also wrong for a retail or consumer brand hoping the lessons transfer. Mostly they do not.
Test question: "How would you explain one of your models to a supervisor who is hostile to it?" This is the actual job in financial services, and people who have done it have a rehearsed answer.
7. The future-of-work architect, for role design rather than reassurance
Good for: workforce planning, organisational design offsites, leadership development. They work on what the job becomes, not whether the job survives: new role definitions, new performance measures, new team shapes.
Wrong for: frontline audiences worried about this quarter, to whom org-design abstraction reads as cold. Send a psychologist into that room first, then the architect three months later. Our guide to workplace transformation speakers has the sequencing.
Test question: "Describe a role that was redesigned rather than eliminated, and tell me what the new job description says." Vague futurism cannot answer that. A practitioner will quote lines from it.
8. The practitioner-founder, for credibility built out of scars
Good for: innovation summits, CEO peer groups, engineering all-hands. They have made the resource calls, hit the data quality wall and handled the anxiety of a team being rebuilt around new tools, and they can say so without a comms filter.
Wrong for: most Fortune 500 rooms, honestly. A forty-person company's playbook does not transfer to a forty-thousand-person one, and founders often do not know which parts break. Also wrong if they are mid-raise, because you will get the pitch.
Test question: "What in your approach would break at ten thousand employees?" A founder who says "nothing" has not thought about your company at all.
9. The transformation lead, for the stalled middle
Good for: programmes eighteen months in with flat adoption, where the pilot worked and the rollout did not. They know the unglamorous causes: unclear ownership, training that was cut to fund licences, middle managers quietly protecting their old process.
Wrong for: kickoff energy. They are deliberately undramatic, which is the point, and which makes them a poor opening act.
Test question: "What did you have to stop doing to make the programme work?" Real transformations have a subtraction list. People who only advised on them talk exclusively about additions.
10. The educator, for the comprehension gap under everything else
Good for: company-wide fluency programmes, management training, and workshop formats rather than keynote slots. Most of the distance between what your tools can do and what your people do with them is comprehension, not motivation.
Wrong for: a headline keynote in front of two thousand people, where teaching turns into a lecture. Also wrong for technical audiences, for obvious reasons.
Test question: "Teach me one thing about how a model works, in ninety seconds, without comparing it to a human brain." Teachers can do this on demand. Presenters reach for the brain analogy immediately.
The questions do more work than the categories
Every one of those test questions exists because we ask versions of them ourselves before a speaker gets in front of a client. Not credentials, not follower counts. What did you ship, what broke, who from a recent audience will take my call, and what would you change for this specific room.
That is also the whole intake. Tell us the four things about your event and we send three to five verified speakers back within 24 hours, with the fee and our 17.5% commission printed on the same line so you can see exactly who gets what. No account, no discovery call, and nothing about how we work that you have to ask twice to find out.
If you only remember one thing: pick the category from the problem, then spend your energy on the test question. The category narrows the field. The question tells you whether the person in front of you belongs in it.