Why this market got weird
Since late 2022, the number of people speaking on AI has gone through the roof. Plenty are great. Plenty spun up a deck in a weekend and started quoting five figures. For an event planner, The risk of having the audience receive misinformation is certainly out there .
1. Light on the details
Anyone can say "transformer" or "agentic." Ask a candidate to explain, in plain English, why models hallucinate, or when you'd use a RAG instead of fine-tuning. If the answer turns into a word salad or a pivot to a client story, move on. Someone who has actually built with these tools can explain the mechanics without hiding behind the jargon.
2. Ask what they've shipped
Real practitioners have receipts: products, internal tools, MCP servers, automations, agents that are used with real clients. Ask what they've built with AI in the last six months. A vague answer about "advising Fortune 500 clients" is a yellow flag. A specific answer about a workflow, a prompt architecture, or why they picked one model over another is a green light.
3. Match the speaker to the room
A room of engineers doesn't need "What is ChatGPT?" A room of board members doesn't need a live demo of building your own MCP server. Most speakers have a natural altitude they operate at, whether that's strategic, operational, or hands-on technical. Be explicit about who's in the seats and what level of talk you are looking for.
4. Look for a balanced take
Watch out for the two extremes. The evangelist who thinks AI solves everything, and the doomer who thinks Skynet is the ievitable conclusion. Both are fun to book and both age badly. The speakers worth paying for hold real optimism about specific use cases and real skepticism about the hype wrapped around them.
5. Watch a full talk before you book
Sizzle reels are built to hide the weak spots. Ask for an unedited recording of a recent full-length keynote if it's available, ideally in front of an audience that looks like yours. Then pay attention to how they handle Q&A. Live questions expose depth faster than any slide can. At the very least, ask for testimonials from the audience and the event managers.
6. Questions to ask on the discovery call
What's something about AI you were wrong about a year ago?
Which tools are you using every day?
What's a claim in the industry right now you disagree with?
How do you tailor a talk to a specific audience?
What do you want our attendees to be able to action on immediately?
Red flags
Bio leans on media logos instead of shipped work.
Every talk sounds the same regardless of audience.
Won't share a full-length recording.
Can't name a specific project they built or led.
Every prediction is either utopian or apocalyptic.