You have budget for one session and a room full of people who have heard about AI for two years and changed almost nothing about how they work. The question is whether to spend it on someone who will stand up and reframe the whole thing, or on someone who will sit down with your team and build something.
That is not a preference. It is a diagnosis, and getting it wrong is the most common way event budgets are wasted.
A keynote changes how a room thinks for an afternoon. A workshop changes what a team does on Monday. Both are worth paying for. They are not substitutes, and a keynote booked to do a workshop's job fails quietly, because everyone enjoys it and nothing happens afterward.
Start from the sentence you want to be true in six weeks
Before you look at a single speaker, write down what should be different six weeks after the session. Then look at what you wrote.
If it is a sentence about belief, book a keynote. "Our regional leaders stop treating AI as an IT project." "Our board understands why we are spending this money." "The organization accepts that this is happening whether or not we like the pace." Belief moves in large rooms, from credible outsiders, in a compressed hour. That is exactly what a keynote is engineered to do.
If it is a sentence about behavior, book a workshop. "Our claims team drafts first-pass responses with the tool and reviews them properly." "Our managers know how to check work their reports produced with AI." "Every product manager has a working prompt library for their own backlog." Behavior does not move because someone was inspiring. It moves because people did the thing once, badly, with help, in a room where it was safe to be bad at it.
If you cannot write either sentence, that is useful information too. It usually means the session is a slot on an agenda rather than an intervention, and a keynote is the honest thing to buy.
The comparison, plainly
| Keynote | Workshop | |
|---|---|---|
| What it changes | What the room believes | What a team does |
| Best group size | 100 to several thousand | 8 to 30 per facilitator |
| Typical length | 30 to 60 minutes plus Q&A | 3 to 4 hours, or a full day, or a series |
| Success looks like | People quoting it a week later | Work products that exist because of it |
| Who should be in the room | Everyone, including people who will never touch the tool | Only people who do the specific work |
| The speaker's core skill | Holding a large room and earning trust fast | Diagnosing a group in real time and adjusting |
| What you take away | A shared reference point | Artifacts, prompts, drafts, a decision framework |
| Typical spend | Wide range by profile, see our fee guide | From $7,500 half-day, from $12,000 full-day |
| Fails when | You needed behavior change | You invited the whole company |
The row that catches most people out is group size. A workshop with 120 people in it is a keynote with worksheets. It feels participatory for the first twenty minutes and then reverts, because a single facilitator cannot circulate, cannot see who is stuck, and cannot pull the room back from a bad tangent. If your headcount is fixed and large, either run parallel tracks with co-facilitators or accept that you are booking a keynote and design it well.
What a good AI workshop actually contains
Ask three facilitators what happens in their workshop. Two will describe a curriculum. One will ask you what your team is currently doing badly. Book that one.
A workshop that is worth the money has these properties.
Hands on keyboards, most of the time. Not hands in the air. The ratio to look for is roughly one third framing and two thirds doing. If the agenda you are sent is a list of topics with a "hands-on exercise" bolted onto the end of each, it is a lecture with breaks.
Their work, not case studies. The exercises should be built from what your team actually ships: your tickets, your reports, your customer emails, your risk assessments. Generic examples are the single biggest reason AI training does not transfer. People do the exercise about the fictional retailer, agree it was interesting, and then never map it onto their own inbox. A facilitator who asks for anonymized samples of your real work before the session is doing the job properly.
Your tools, in your environment. If your organization has standardized on a particular assistant or has a locked-down internal deployment, the workshop should run inside it, with your permissions and your restrictions in place. A session run on a consumer tool your people cannot legally use at work teaches a skill they cannot practice.
Room for mixed exposure. In any real team you will have two people who have built their own automations and five who have never opened the tool. A good facilitator plans for that spread rather than being surprised by it, usually by pairing across the gap and giving the advanced participants a harder version of the same task. Ask directly how they handle it. Vague answers here are a reliable warning sign.
A leave-behind that keeps working. Prompt libraries, checklists, a review standard for AI-assisted work, the automation somebody actually built before the day ended. The value should compound after the room clears. A deck does not compound.
A named outcome, agreed before booking. Every workshop we place starts with a scoping call, and the agenda is designed backward from the outcome you named rather than forward from a standard curriculum. The full catalog of formats, from a 90-minute board briefing through to train-the-trainer programs, is on our workshops page with the starting fees printed.
How to brief for a workshop
A keynote brief describes an audience. A workshop brief describes a task. That is the whole difference, and it is why briefs written for one format land badly when reused for the other.
Include these, in writing:
- The behavior. What specific work should be done differently, by whom. Not "improve AI literacy." Something like "our four analysts should be producing first-draft variance commentary with the tool and reviewing it against a standard we agree in the room."
- Current state, honestly. Who has used what, how often, and what went wrong last time. If a previous rollout stalled, say so and say why you think it stalled. Facilitators design around resistance, and they cannot design around resistance you hid.
- Constraints. Which tools are approved, what data cannot leave the building, what your legal and security teams have already ruled out. Half of a bad workshop is a facilitator teaching something your policy forbids.
- Headcount and homogeneity. Twelve people from one function is a different session from twelve people from nine functions. Both work. They are not the same design.
- Time and follow-through. Whether this is a one-off or the first of a series, and who inside the organization owns what happens next. Workshops with no internal owner decay fastest.
Then send it. The rest of the process is identical to booking any speaker, and the mechanics are in our step-by-step booking guide. Four things about your session gets you verified names back within 24 hours through get a shortlist, with no account and no discovery call first.
Facilitating is a different job from speaking
This is the part organizers underestimate. A brilliant keynote speaker is not automatically a good facilitator, and the failure is not obvious until you are two hours into a day you cannot restart.
A keynote is a performance with a fixed shape. The speaker controls the narrative from the first line to the last, absorbs the room's energy, and is judged on whether the argument landed. Interruption is a risk to manage.
Facilitation is the opposite. The facilitator's job is to hand control away and then keep the session from falling apart while they are not holding it. They have to notice that table three has gone quiet, that one participant is quietly dominating, that the exercise everyone is nodding along to is not actually working, and that the plan for the next ninety minutes should change. They have to be comfortable being less interesting than the work.
The tells are easy to check for. Ask a candidate what they cut from their standard day and why. A facilitator has cut things, recently, in the room, and can tell you what they dropped when a group turned out to be further behind than the brief suggested. Ask what they do when half the room finishes an exercise in ten minutes and half is still stuck at twenty-five. Ask for a reference from someone who sat through the whole day rather than someone who booked it.
We screen for this specifically, because the two skills are not interchangeable and a lot of people market themselves for both. Anyone can claim to be an AI expert. We check, and for facilitators we check a different set of things than we do for keynote speakers. That process is written up in how we verify a speaker.
Booking both, in the right order
The strongest programs use both, and the sequence matters. Keynote first, to the wide room, to establish the premise and give people permission to take it seriously. Workshops second, to the specific teams, within a few weeks while the keynote is still a live reference.
Reversed, the workshop teams go back to colleagues who have not had the framing and lose momentum arguing for it. Separated by six months, the keynote has become a thing that happened at last year's offsite.
If the underlying goal is a durable change in how people work rather than a single event, that is a program rather than a booking, and it is worth reading what to look for in a workplace transformation speaker before you commit to either. And if budget is the constraint that is forcing the choice, our 2026 fee guide sets out the keynote bands so you can compare the two honestly, with our 17.5% commission shown on the same line as the speaker fee rather than buried in the total.