Imagine this scene
A VP has a board presentation in three days. You have the story outlined but need to produce slide copy, a speaker script, and separate speaker notes. These are three different documents that serve three different purposes. AI can produce all three once you understand the distinctions.
This is the most common mistake in presentation work: treating the script and the slide copy as the same content in different containers. They’re not. Slide copy should be sparse — the minimum words needed to anchor the audience while the speaker talks. The script is what the speaker says, often much longer than what’s on screen. Speaker notes are the hybrid: brief cues and key points written for a speaker who knows the material and doesn’t want to read verbatim. AI can produce all three — but only if you ask for them separately and brief them separately.
Key Insight
Slide copy should be the anchor, not the content. If every word the speaker says is already on the slide, the audience stops listening to the speaker.
Most source material for presentations is prose — memos, reports, briefs. AI can translate prose to slide language reliably: give it a paragraph, ask it to produce a headline (5–8 words) and 3 supporting bullets (maximum 10 words each). The output will almost always need editing for voice and precision, but the structural compression is done. This translation process, which takes a writer significant time, takes AI seconds.
Not every speaker wants a full script. Some find scripts constraining and perform worse when they feel like they’re reading. For experienced speakers, talking points are often more useful: the 3–4 key things to say per slide, in a logical order, with the freedom to express them naturally. For less experienced or high-stakes presentations, a full script gives the safety of having every word prepared. Know your speaker and ask AI for the right output format.
Speaker notes occupy a specific middle ground: they’re not the script and not the slide copy. They’re the prompts a speaker uses during delivery. They should remind the speaker of: the key point they want the audience to leave with from this slide, any specific statistic or anecdote they want to include, and any transition into the next slide. Ask AI to write speaker notes separately from both the slide copy and the script — with a distinct instruction: “Write these as brief cues the speaker will glance at, not paragraphs they’ll read.”
Produce slide copy, talking points, and speaker notes for a real or hypothetical presentation section using AI.
Choose a 5-slide section of a presentation: an announcement, a strategy update, a results summary, or an issue briefing. Write or identify the source material — the key story you want to tell in those 5 slides.
Ask AI to produce slide copy for each slide: a 5–8 word headline and 3 supporting bullets of no more than 10 words each. Review each slide for density — if a bullet requires a full sentence to be meaningful, it’s too long.
Now ask AI to write talking points for the same 5 slides — 3–4 things the speaker should say per slide, in a conversational rather than scripted style. Compare the talking points to the slide copy: are they genuinely different layers?
Finally, ask AI to write speaker notes for each slide: brief cues only, not full sentences. Include: the key takeaway for this slide, one specific detail to mention, and the transition to the next slide.
Read the whole set aloud as if presenting. Note where the slide copy, talking points, and speaker notes work together smoothly — and where the layers collapse into repetition. Edit until each layer is doing its own distinct job.
