Speaker notes that read like a script become exactly that — a monotone read-aloud. Good notes prep the speaker with cues, key phrases, and what to leave out.
Who this is for
Anyone presenting (founders, IC engineers, salespeople), comms teams supporting executives, course / training facilitators.
When not to use these prompts
Don’t use these for slides that already say everything (those are handouts). Don’t use them as full scripts unless the speaker is very inexperienced.
Prompt anatomy / structure formula
Every prompt should carry six elements:
- Role: who AI plays — analyst, chief of staff, manager.
- Context: team / org / project / data scope.
- Goal: one deliverable — table, doc, talking points, plan.
- Constraints: word count, must-include fields, audience seniority.
- Tone: confident, neutral, factual — depends on audience.
- Examples: 1-2 samples of prior work to anchor format.
Best for
- Cue-based speaker notes
- Per-slide talking points
- Anticipated-question prep
- Transition cue notes
- Timing notes
12 copy-ready prompt templates
1. Cue-based notes (not scripts)
Slide: `{slideTitle}`. Slide content: `{content}`. Write speaker notes as: 3-5 bullets of key phrases + 1-2 transition cues. Skip full sentences. Total ≤ 60 words.Variables to swap: slideTitle, content
2. Per-slide opener / closer
For each slide, give: (a) Opening sentence (≤ 10 words), (b) Key message (≤ 15 words), (c) Closing transition to next slide (≤ 10 words). Total: 3 lines per slide.3. Anticipated-question prep
For this deck, predict 5 questions audience will ask. For each: which slide will prompt it, 2-sentence answer, fallback if pressed. Don't bury answers in slides — speaker handles.4. Transition cue notes
Audit my deck for awkward transitions. For each slide pair: write a one-line bridge sentence the speaker says ("So far we've seen X. Next, we look at Y."). Output as a transition table.5. Timing notes per slide
My talk is `{minutes}` minutes, `{nSlides}` slides. Allocate time per slide (some get more): output a timing plan. Flag slides that are too dense for their time slot.Variables to swap: minutes, nSlides
6. Energy / pace cues
Add pacing cues: where to pause (after a hard number), where to accelerate (recap slides), where to slow down (a counter-intuitive claim). Output as cues on each relevant slide.7. What NOT to say
Audit my deck for things speakers commonly over-share that hurt: (a) caveats that erode confidence, (b) jargon for general audience, (c) tangents. List per slide.8. Demo-fail backup
My deck includes a live demo at slide `{n}`. If it fails, what does the speaker say? Write 3 backup lines: (a) acknowledge & continue, (b) screen-share alternative, (c) "we'll get to this offline" pivot.Variables to swap: n
9. Story-anchored speaker notes
For data-heavy slides, add one story anchor: a real customer, a real moment, a real number. Speaker says story before the data, not after.10. Per-slide one-word summary
Add a one-word summary to each slide's speaker note ("Problem", "Hook", "Proof", "Ask"). Helps the speaker re-orient mid-presentation.11. Q&A close cue
Write the speaker's closing line + Q&A open: (1) recap headline takeaway in 1 sentence, (2) the one thing to remember, (3) Q&A invite. Avoid "Any questions?" — too low-engagement.12. Speaker-note hygiene
Audit my speaker notes: (1) Any notes that look like a script (full paragraphs)? (2) Any slide with no notes (mistake or intentional)? (3) Any slide where the note re-states the slide? Output a cleanup list.Common mistakes
- No specific context — output is generic.
- Skipping fact-check — AI invents numbers when given soft inputs.
- Vague audience — output overshoots or undershoots seniority.
- No word limit — receivers won’t read past line 5.
- Same template for every situation — readers tune out.
- No “decision needed” framing — readers don’t know what to do.
- Forgetting to attach the source data — claims without receipts.
How to push results further
- Always specify audience level (IC / Manager / VP / CEO).
- Cap length: 1-page max for tactical, 3-bullet for executive.
- Lead with the ask / decision needed. Context after.
- Attach source data link — saves a follow-up email.
- Read aloud before sending; cut every sentence > 25 words.
- Use AI to draft and audit, not to ship without review.
- Save best examples; reuse format, refresh content.
Practical depth notes
Use these prompts as starting points, not final answers. For Presentation Speaker Notes Prompts That Aren’t Read Aloud, the useful extra work is to replace every generic placeholder with a real constraint: audience, channel, length, brand voice, examples to imitate, and examples to avoid. Run at least two versions with different constraints, then compare the outputs side by side instead of accepting the first polished response.
A good result should pass three checks: it is specific enough that another person could reuse it, it avoids vague praise or filler, and it gives you an editable artifact rather than a broad suggestion. If the output feels generic, add one concrete reference, one forbidden pattern, and one measurable success criterion before rerunning the prompt. Before saving a prompt as reusable, test it on one realistic input and one edge case. The realistic input proves the template can produce the normal deliverable; the edge case shows whether it handles messy constraints, missing context, or an unusual audience. Keep the better output, but also keep the failed version with a note on what was missing. That small failure log is what turns a prompt collection from a list of nice sentences into a practical working library.
FAQ
- How long should this doc be?: Match audience patience. Tactical: 1 page. Executive: 3 bullets + link.
- Can AI replace the analyst?: For first drafts and templates yes; for judgment calls no.
- How often refresh?: Cadence-driven (weekly / monthly / quarterly) but adjust when audience signals fatigue.
- Should I include risks?: Always. Pretending no risk exist erodes trust on the next update.
- How to keep it fact-checked?: Attach data sources and have a peer skim numbers before sending.
- Can AI generate the data itself?: No — AI invents plausible numbers. Connect to real data sources.