Having coached presenters for more than fifteen years, I’ve found that a truly persuasive presentation rarely comes together perfectly on the first try. The ones that truly energise a room usually go through a lot of refining: they are rewritten, tested with real audiences, and adjusted, cutting parts that lose people and keeping those that resonate. This valuable work is done through genuine rehearsals with real people, not in just one quick pass.
I wanted to see what happens when you skip straight to the finished version without any of that.
I picked climate change deliberately. It’s a topic where the facts are settled and the science is genuinely overwhelming, which makes it the fairest possible test: if AI-generated slides can’t move a room on a subject this well-evidenced, volume of data was never going to be the thing that was missing.
I asked AI for a slide deck to persuade an audience that climate change is real, urgent, and needs action now. Fourteen slides came back: scientific consensus, temperature rise, CO₂ levels, ice melt, extreme weather, wildfires, ocean acidification, human impact, economic cost, UK-specific figures, future projections, the case for renewables, a closing call to action. Every figure sourced: IPCC, NASA, NOAA, WMO, the Met Office. On the surface, it read like a rigorous briefing.
The Deck, Exactly as AI Wrote It
Slide 1 — Title
- Climate Change: The Evidence, The Urgency, The Action — A data-driven analysis of the state of the climate.
Slide 2 — The Scientific Consensus
- 97% of climate scientists agree that human activity is warming the planet.
(IPCC, NASA)
Slide 3 — Global Temperature Rise
- Earth’s average surface temperature has increased 1.1°C since the late 19th century.
- The last 10 years were the warmest decade on record.
(NASA, WMO)
Slide 4 — Atmospheric CO₂ Levels
- Pre-industrial CO₂: 280 ppm
- Today: 424 ppm (2024 measurement)
- Highest in at least 800,000 years
(NOAA, NASA)
Slide 5 — Ice Melt & Sea Level Rise
- Greenland is losing ~270 billion tons of ice per year.
- Antarctica is losing ~150 billion tons per year.
- Global sea level has risen 20 cm since 1900.
- Rate of rise has doubled since the 1990s.
(NASA, IPCC)
Slide 6 — Extreme Weather Increase
- Heatwaves now occur 3x more often than in the 1980s.
- Major hurricanes have increased in intensity by ~25%.
- Flooding events in Europe have increased by ~50% since 1980.
(WMO, Met Office)
Slide 7 — Wildfires
- The area burned by wildfires globally has increased by ~75% since the 1970s.
- 2023 Canadian wildfires released 1.7 billion tons of CO₂ — more than Canada’s annual emissions.
(WMO, Copernicus)
Slide 8 — Ocean Warming & Acidification
- Oceans absorb 90% of excess heat from global warming.
- Ocean heat content is at its highest level ever recorded.
- Ocean acidity has increased by 30% since the Industrial Revolution.
(NOAA, IPCC)
Slide 9 — Human Impact
- Climate-related disasters displaced 32 million people in 2022.
- Heatwaves cause ~500,000 premature deaths per year globally.
- Air pollution kills 7 million people annually.
(UNDRR, WHO)
Slide 10 — Economic Impact
- Climate disasters cost the world $313 billion in 2022.
- By 2050, climate change could reduce global GDP by up to 18%.
- Every £1 spent on climate adaptation saves £4–£10 in avoided damages.
(ND-GAIN, Swiss Re, UK Treasury)
Slide 11 — UK-Specific Evidence
- The UK’s average temperature has risen 1.3°C since 1884.
- 2022 was the UK’s hottest year on record.
- Sea levels around the UK have risen 16 cm since 1900.
(Met Office, UK Climate Projections)
Slide 12 — Future Projections (IPCC AR6)
- If warming reaches 2°C: extreme heat events become 4x more frequent.
- Sea level rise could exceed 1 metre by 2100.
- Up to 3 billion people could live in areas too hot for human survival.
(IPCC AR6)
Slide 13 — The Opportunity
- Renewable energy is now cheaper than fossil fuels in most regions.
- Solar costs have fallen 89% since 2010.
- Wind costs have fallen 70%.
- Every £1 invested in clean energy returns £3–£8 in economic benefits.
(IRENA, BloombergNEF)
Slide 14 — The Call to Action
- The evidence is overwhelming. The risks are escalating. The window for meaningful action is closing.
- The question is no longer “Is climate change real?” It’s “How fast can we act?”
Read straight through; it’s thorough and well-sourced. It’s also thirty-odd facts with nobody standing behind any of them, which is exactly the problem the rest of this piece is about.
Statistics Without Feeling Rarely Move Anyone
Count them, and the AI deck features over thirty distinct statistics spread across fourteen slides: 424 parts per million, 270 billion tons of ice annually, 32 million displaced people, 500,000 heatwave-related deaths, and 7 million deaths from air pollution. Most of them are broadly accurate and linked to reputable climate sources, at least as far as the citations AI attached to them go; I haven’t independently re-verified all thirty-odd against primary sources myself. Even taken at face value, plenty of people would find the sheer weight of it compelling on its own. Compelling and moving aren’t the same thing, though. What this much data rarely does is drive the kind of urgent, sustained action the numbers actually call for. We’ve grown a little immune to statistics like these; there are simply too many of them, in too many decks, for any single number to keep landing with force.
The psychologist Paul Slovic has spent much of his career studying how compassion changes as the number of people involved increases. In one well-documented study, people who read about a single named child facing hunger gave more money than people who read a statistical description of the same famine. Adding the statistics alongside the child’s story didn’t help either; it measurably pulled donations back down. None of this is a fixed law, and not every attempt to reproduce it succeeds, but the pattern shows up often enough to take seriously: past a certain point, a number stops reading as a collection of individual people and starts reading as an abstraction.
What separates a data-heavy presentation that works from one that doesn’t isn’t the amount of data. It’s whether each fact has been chosen for a reason, connected to something the audience already understands or feels, and tested on real people rather than just assumed to land. AI’s thirty-odd statistics haven’t been through any of that. They’re accurate, but nobody has actually chosen them for this room.
I’ts worth noting that not everyone in that room would respond the same way. Psychologists call the relevant trait need for cognition, and people high in it, the ones who genuinely enjoy working through an argument, are measurably more persuaded by strong, dense factual material than by anything wrapped in narrative or feeling. For a room of engineers, actuaries, or research scientists, AI’s raw deck might do real work almost as it stands. The point was never that data can’t persuade anyone. It’s that most rooms aren’t that room, and knowing which one you’re facing, before you decide how much to lean on feeling versus fact, is exactly the audience-reading job Resonance already asks of you.
The Line That Gave It Away
Slide 14 closes with this: “The evidence is overwhelming. The risks are escalating. The window for meaningful action is closing. The question is no longer ‘Is climate change real?’ It’s ‘How fast can we act?’”
Read it back. Three short, escalating statements, then a reframed rhetorical question. It’s a recognisable shape; the kind of line that looks punchy on the page but sounds hollow the moment it’s spoken aloud. Worse, it asks the audience to think rather than feel, at exactly the point when a persuasive close needs to do the opposite. Everything before it in the deck was already analysis. This was the one place built to move people, and it reached for reasoning instead of feeling, which risks undermining the very impact it’s going for.
The Question It Couldn’t Survive
Imagine the presentation is finished, and someone in the audience asks the natural follow-up: “Okay, so what do you want me to actually do?” As you review all fourteen slides, you find no clear answer. There is ample evidence that the problem exists, but nothing that directs a specific individual in a specific setting on what to do differently on Monday morning.
Presenters who test material on real audiences tend to get asked exactly this kind of question, and often end up revising their material because of it. AI generated its response once, without an audience, and never faced a follow-up at all. To be fair, a savvy presenter reviewing this deck before ever stepping in front of anyone would likely spot the missing call to action themselves and fix it in rehearsal, no live audience required. The deeper issue isn’t that AI could never produce a deck with this gap in it, it’s that AI itself has no way of noticing the gap. It has no audience to disappoint and no habit of asking “so what do I actually want them to do.” That noticing is still a human job, whether it happens in rehearsal or in the room.
One Slide in Fourteen
Count the slides again, this time by what job each one is doing rather than what it says. Thirteen of AI’s fourteen slides build the case that the problem is real, severe, and getting worse. Exactly one, Slide 13, makes the case that anything is actually working. That’s not a minor imbalance. It’s the same structural failure researchers behind climate messaging keep flagging: threat that isn’t matched by a genuine, felt sense that action works tends to produce numbness or disengagement, not urgency.
It gets worse on closer inspection. When Slide 13 does show up, it gets exactly the same treatment as every slide before it: three flat bullet points, no more craft, no different register than the wildfire slide or the death toll slide. AI had no way of knowing this particular slide needed to sound different, because it has no sense that different content should land differently in the first place. Thirteen slides of dread and one slide of proof, delivered in identically flat tone, isn’t balance. It’s an accident of how the facts happened to sort themselves into two buckets.
How to Fix It
Worth saying plainly before going further: these three facts are devastating on their own, and accurate. The problem was never that they’re unconvincing. It’s that we’re handed big numbers like this so often, in so many decks and headlines, that they’ve become hard to actually feel, however true they are.
Take Slide 9, exactly as AI wrote it:
Climate-related disasters displaced 32 million people in 2022. Heatwaves cause approximately 500,000 premature deaths per year globally. Air pollution kills 7 million people annually.
I ran it through the six layers of The Presenter’s Operating System™, the model I use with clients. Here’s what each layer actually does to material like this, and it’s worth noticing that every one of them is something I’ve watched the strongest presenters do instinctively, long before anyone gives it a name.
The Foundation: Headstuff
Before any of the six layers, there’s Headstuff: the older, unrelated stories you already carry into a room like this one, things like “if I let people see how much this actually scares me, they’ll stop taking me seriously,” or, on a topic like this one, “some people in this room will never believe this no matter what I say, so why bother trying,” or “we’ve been raising this alarm for years and nothing’s changed, so what’s the point.” Those last two are worth naming honestly, because they rarely stay quiet. Left unnamed, they tend to turn a presenter defensive, preachy, or quietly resigned before they’ve said a word, writing off part of the room, or writing off the whole effort, instead of actually trying to reach it. Naming the belief isn’t enough on its own; it needs somewhere to go. Write it down on paper before you start building anything, put it outside yourself rather than carrying it silently, and make a deliberate choice to build this presentation as the person who actually cares about reaching this room, not as the person defending against being seen to care. AI has no version of this layer. It has no history to set down, which is exactly why its output can sound confident without ever sounding lived-in.
1. Message
Slide 9 gave you three separate facts: 32 million displaced, 500,000 heatwave deaths, 7 million deaths from air pollution. That isn’t a message. It’s raw material sitting in a queue, unranked and unresolved. A message has to be one clear, concise, compelling sentence that everything else in the presentation serves. Something closer to: climate change isn’t a future problem, it’s already displacing tens of millions of people and killing millions every single year. Once you have that sentence, every fact either earns its place by supporting it or it gets cut.
Sharpen it further and it’s short enough to say in one breath, and to actually share:
| Climate change isn’t a future problem. It already displaces 32 million people, kills 500,000 through heat, and kills 7 million through dirty air — every single year. The question was never if. It’s how fast we act. |
That last line is doing something worth noticing. It’s built on the exact rhetorical question AI’s own Slide 14 tried and failed with: “the question is no longer if, it’s how fast.” On Slide 14, that question had nothing behind it, so it sounded hollow. Here, it’s sitting on top of three real, specific numbers the room has just heard. Same question, but earned this time instead of asserted. That’s the actual difference between raw Message and everything that came before it in the deck: not cleverer phrasing, just something real underneath it.
2. Intention
Intention means deciding, specifically, what you want this audience to feel, why you want them to feel it, and exactly how you’re going to get them there. Not “concern,” in the abstract, and not guilt either; in my experience, guilt tends to make people defensive rather than active. Something closer to: I want this room to feel the same protective pull they’d feel about someone they love, not the distant unease they feel about a stranger in a headline. That’s the difference between an audience that nods along and one that actually moves. I’ll get there with an image, not a statistic.
| THE TARGET FEELING
Picture someone you love — a parent, a child, whoever comes to mind first — caught in whichever version of this reaches them first: heat so severe they’re struggling to think straight and there’s no shade for miles, smoke so thick every breath catches in their throat, water rising fast enough that they’ve got ten minutes to leave, not ten hours. Pick whichever one unsettles you most. That discomfort, that urge to go get them out of it — that’s the feeling this needs, not a decimal point in a report, not guilt about a graph. |
That’s what this layer is actually deciding: not a topic, a target feeling, chosen on purpose, aimed somewhere specific. Everything from here on, the number you pick, the image you build it into, is in service of landing that one feeling, not the next one that happens to occur to you.
3. Resonance
Resonance is really about how well you know your audience, specific enough that material lands as theirs, not just as accurate. In this case, that bar is lower than it sounds. You don’t need to know much about the room to know they have people they love, and that most of them have some frame of reference for a full plane, whether they fly often or not. Those two things are close to universal, which is exactly why they work as the anchor for the protective feeling Intention already named: the pull you’d feel picturing someone you love in that heat, that smoke, that rising water. Resonance’s job here is turning that feeling into a concrete image people can actually hold onto, not a bare number. Five hundred thousand heatwave deaths a year works out to roughly 1,370 people a day, close to the number of passengers on three or four full jumbo jets. Nobody would accept that many plane crashes a year without grounding every airline on earth. It’s happening anyway, just spread out and uncounted the same way. Slide 9 never attempts this kind of translation; it reports the total and moves straight on to the next statistic. A more specific audience could go sharper still. For a general one, this is already the right choice.
4. Clarity
Clarity means checking everything you’re about to say or show against one question: does it serve the message? The other two facts, 32 million displaced, 7 million air pollution deaths, are just as shocking on their own, and they’re not being thrown out of the presentation. This is about this slide, not the whole talk. Resonance already did the hard work of turning 500,000 into something people can picture, and stacking the other two on top doesn’t strengthen that image, it just competes with it. So for this slide, they go, and earn their own moment elsewhere instead, with the same care put into making them land. Ask the same question of everything left here: does this serve the message? If not, cut it from this slide, however true it is.
Giving the Other Two Their Own Moment
Clarity said these two facts would earn their own moment elsewhere in the deck instead of being cut for good. Here’s that moment kept: what displacement and air pollution look like once they get the same translation, on their own slides.
| On screen | Spoken aloud |
| 32 million displaced in 2022
≈ 1 person, every second, all year |
“32 million people displaced by climate disasters in a single year works out to roughly one person, somewhere in the world, forced from their home every single second.
By the time I finish this sentence, that’s another six or seven people who no longer have anywhere to sleep tonight.” |
Image concept: a suitcase packed and abandoned outside a front door.
| On screen | Spoken aloud |
| 7 million air pollution deaths a year
≈ 1 person, every 4–5 seconds |
“Air pollution kills about seven million people a year, roughly one person every four or five seconds, day and night, somewhere on earth.
In the time it’s taken you to read this slide, several more people have died from something they breathed, not something they chose.” |
Image concept: a child using an inhaler outdoors in hazy, polluted air.
5. Centring
Centring has nothing to do with pacing. It’s about connecting with yourself before you can attempt to connect with anyone else, and it needs an actual practice, not just an instruction to “connect.” In my experience, your emotional state doesn’t stay a private experience once you’re standing in front of a room; it tends to become the room’s emotional climate, whether you mean it to or not. Before you say a word, use a 5–2–8 reset: inhale for five seconds, hold for two, exhale for eight. It’s fast, discreet, and it steadies your nervous system enough to walk in with real composure rather than performed composure.
Ground yourself at the same time. Feet hip-width apart, weight settling down through them rather than staying up in your shoulders. Knees unlocked, not braced. Spine lengthening because you let it, not because you force it. Chest open, unguarded, not puffed out. A moment of stillness, not for effect, just to arrive. None of that is decoration. A voice with nothing solid underneath it has to work for authority; a voice standing on a grounded body doesn’t.
Then pause for two or three seconds before you speak, and let the room settle around you instead of launching straight into the material. Walk in still rushed and unsteady, and the audience will sense the gap even if they couldn’t name what caused it. AI skips this layer entirely, because there’s no body to ground and no nervous system to steady in the first place.
6. Voice
Voice is everything that comes out once the first five layers are actually done: the words themselves, and just as much, how you say them and how you hold yourself while you say them.
Practically, that means recording yourself saying the line below back and listening for monotony: does the plane comparison get any different treatment than the sentence before it, or does everything come out at the same pitch and pace? A deliberate pause before “gone,” a slightly lower pitch on the final line, a touch more pace through the middle section where you’re building the picture; in my experience coaching this, that’s not decoration, it’s doing a good deal of the persuading. Then check what your body’s doing while you say it: arms crossed, weight shifting, eyes on your notes instead of the room. The leadership researcher Carol Kinsey Goman has made the point that how a message is delivered, vocally and physically, can undermine even a genuinely strong script, and most speakers underestimate by how much.
AI’s deck never leaves the page. Every slide is written to be read, not spoken, so Voice never actually happens in it. Voice only exists once someone stands up and says the words out loud, carrying everything from the five layers before it. Here’s what that sounds like, once it actually has been said:
| On screen | Spoken aloud |
| 500,000 heatwave deaths a year
≈ 4 full passenger planes, every day |
“Heatwaves kill about 500,000 people a year. Spread that across the calendar and it’s roughly four full passenger planes, every single day, with nobody walking away.
If that were happening in aviation, we’d have grounded every airline on earth by lunchtime. It’s happening anyway. We just don’t count it the same way.” |
Image concept: a single empty airline seat with the seatbelt still fastened.
Same underlying fact; one image instead of three numbers, and a comparison the room can actually hold onto after the slide’s gone. That’s the distance between what AI handed me and what I’d actually be willing to stand up and say.
What Every Slide Could Feel Like
That’s one slide. Here’s a first pass at doing the same thing to the rest of the deck: not a full six-layer treatment for each one, just the single translation that would matter most, going from a number nobody can picture to an image somebody could actually carry out of the room. I’ve also suggested one image concept per row — these decks are currently just text and bullets, and that’s part of the problem this whole piece is about.
| AI’s version | What it could feel like | Image concept |
| 97% of climate scientists agree. | If 97 out of 100 firefighters told you the building was on fire, would you wait for the other three to agree before you left? | Firefighters battling a raging house fire |
| Global temperature up 1.1°C since the 1800s. | A 1°C fever is enough for a doctor to take you seriously. This is that, for the whole planet. | Hospital heart monitor flashing a red critical alert |
| CO₂ at its highest in 800,000 years. | 800,000 years predates our species. No human being has ever breathed air like this. | Scientist extracting an ice core sample |
| Greenland losing ~270 billion tons of ice a year. | Roughly 3 Olympic swimming pools of meltwater, every second, all year round. | Massive glacier collapsing into the sea |
| Heatwaves 3x more frequent than the 1980s (AI’s figure; a 2021 peer-reviewed study found concurrent large heatwaves seven times more frequent, a related but not identical measure). | Either way: one scorching summer in the 1980s, several most years now. | Deserted playground baking in extreme heat |
| 2023 Canadian wildfires released 1.7 billion tons of CO₂ (AI’s figure; more recent WRI and peer-reviewed research puts the real total closer to 3 billion tons). | More CO₂ in one fire season than every car currently on America’s roads produces in a year, twice over. | Thick wildfire smoke turning a city sky orange |
| Oceans have absorbed 90% of the extra heat. | Equivalent to between 1.5 and 12 Hiroshima-sized atomic bombs of heat going into the sea every second (the figure has climbed as warming accelerates), based on published ocean heat content data — see note below. | Steam rising off the ocean at sunrise |
| 500,000 heatwave deaths a year. | Roughly four full passenger planes, gone, every single day. | Rows of empty seats inside an airplane cabin |
| $313 billion in climate damage in 2022. | That’s roughly $858 million a day, which is more than the NHS in England spends in a day, on everything, once its own annual budget is divided the same way. | Ambulance with lights flashing outside a hospital at night |
| 2022 was the UK’s hottest year on record. | It’s the year London Fire Brigade had its busiest day since the Second World War, with 41 properties destroyed, the same day Britain hit 40°C for the first time in its history. | Burnt-out, fire-damaged house |
| Up to 3 billion people in unlivable heat by 2100. | Close to the entire population of India, China and the United States combined, living somewhere too hot to survive. | Aerial view of a massive crowd |
One gap worth flagging: AI’s claims that hurricanes are up roughly 25% and European flooding up roughly 50% since 1980 don’t match any single authoritative source I could find. Real research exists on both, it just doesn’t collapse into the clean number the deck implies, so treat those two as unverified rather than confirmed. The wildfire figure, by contrast, is verifiably off; real research puts 2023’s Canadian wildfire emissions at roughly 3 billion tons, not the 1.7 billion AI’s deck claimed. The rest is arithmetic anyone can redo.
Use AI for what it did well here: a fast, well-sourced first pass with the scale of the problem laid out correctly. Then do the part it can’t do, the part that only ever comes from real rehearsal in front of real people. Pick one number and make it human. Say plainly what you want the audience to do differently, not just how you want them to feel. And close on something other than a rhetorical question, because a room that’s just sat through thirty statistics needs one thing to actually hold onto, not a cleverer way of restating the problem.
One more thing worth doing, once the rest is in place: open and close on the same image, then let the ending escalate it. This deck could open with “picture someone you love”, then close by sending the room back to that same person and multiplying them by 32 million, the actual number displaced this year. Not guilt. Scale, built on top of a person the room already feels something for, rather than a number they never did. The ending doesn’t introduce anything new. It completes a thought the room has been carrying since the first slide, which is usually what makes a close feel inevitable rather than tacked on.
If it were me delivering this, I wouldn’t leave that person generic on the slide either. I’d put up an actual photograph of someone I love. You may want to consider doing something similar.
Nobody gets years of rehearsal on most presentations. But every one of the six layers above is available on attempt one, if you actually do the work AI can’t do for you.
The Slide That Deserved the Same Treatment
There’s a fourteenth slide sitting quietly in the middle of the deck, the one titled The Opportunity. Renewable energy now cheaper than fossil fuels in most regions. Solar costs down 89% since 2010. Every £1 invested in clean energy returns £3 to £8. It’s just as flat as everything else AI wrote, three bullet points with nothing behind them, and it’s easy to skim past on the way to the more dramatic slides either side of it. That’s a mistake. This is arguably the most important slide in the whole deck, and it never got the treatment Slide 9 did.
Message: renewable energy now beats fossil fuels on cost, in most of the world, already. Not a projection. Not a hope. Already true, right now.
Intention: this is the one moment in the whole talk that isn’t asking the room to feel afraid. It’s asking them to feel capable, that acting isn’t symbolic, it’s already winning on pure economics, with a track record to prove it.
Resonance: the same translation the rest of this piece has been doing, applied to proof instead of threat. Installing a home solar system that cost around $7.50 per watt in 2010 costs about $3 per watt today. That’s not activism, that’s just what happened. And solar isn’t the only proof this works at scale. In 1987, nearly 200 countries signed the Montreal Protocol to phase out ozone-depleting chemicals, hit 98% compliance, and the ozone layer is recovering on schedule. Closer to home: UK coal-fired electricity went from 40% of the grid in 2012 to zero by October 2024, the first G7 country to get there, cutting power sector emissions by three quarters along the way. None of that happened because enough individuals felt frightened enough. It happened because collective action, backed by policy, was pointed at a specific target and followed through.
Clarity: cut everything except the two numbers that prove the point. Solar’s cost collapse. Coal’s actual, completed phase-out. Everything else is detail.
Centring: this is the one slide in the whole deck that should sound different in the room. Not urgent. Steady. This is the moment the audience gets to exhale, and rushing it wastes the only real relief they get.
Voice, spoken the way it would actually land: “Everything I’ve just told you sounds enormous, because it is. But we’ve done this before. Britain ran on coal for forty per cent of its power in 2012. By 2024, that number was zero, and nobody had to be frightened into it. It happened because it started making economic sense, and enough people backed it once it did. That’s not a hope. That’s a track record.”
This is why it belongs here, not as an afterthought: a case for urgent action that never shows the audience proof that action works isn’t more rigorous for leaving it out. It’s just doing the thing the research on doom communication warns against, dressed up as seriousness.
The One Thing Worth Doing Today
The single most useful thing you can do with everything above isn’t really about climate at all. It’s the next AI-generated deck that lands in your inbox, whatever the topic happens to be. Before you present it, run it through the same test: pick one number, build it into an image, say it in your own words, out loud, before anyone else does. That’s the actual point of this piece, applied the next time you need it, which for most people reading a presentation skills blog, will be soon.
Take This Further
If you’ve got a persuasive case to make and only one shot to land it, it’s worth working through the material with someone else in the room.
Explore our presentation skills training for leadership teams.
Join a public speaking course to build these habits from the ground up.
Or work through your own material directly with one-to-one coaching.
Share This
If this changed how you’ll look at the next AI-built deck you sit through, or the next one you’re tempted to build yourself, share it with someone who’s about to hand their presentation over to a prompt and walk away.
