Blog, AI and tools, 6 min read
AI generated decks: what they get right, and where they fall over
Gamma, Tome and the rest are genuinely useful. Here is what they are good at and what still needs a person.
We redesign a lot of AI generated decks, so we see what these tools produce more than most people do. They are not a con. They are genuinely good at some things and reliably bad at others, and knowing which is which saves you money.
What they are good at
- Getting from a blank page to a draft in ten minutes. That first structure is worth something even if you throw half of it away.
- Internal decks. Weekly updates, team readouts, anything where nobody is judging you on the design.
- Consistent formatting across a lot of slides, which is more than most people manage by hand.
- Filling in the obvious sections you forgot to include.
If you are making a deck for Monday standup, using one of these and moving on is the correct decision. Paying a designer for that is a waste of money.
Where they fall over
The first problem is sameness. These tools have a house look, and by now a lot of people have seen it. When you open a deck and think "this was made with a tool", that thought arrives before any of your content does.
The second is the argument. A generated deck will give you a Problem slide, a Solution slide and a Market slide, because those are the sections decks have. It will not tell you that your problem slide is describing an inconvenience rather than a pain, or that your market number is bottom up when it needs to be top down. That judgement is the job.
The third is charts. Generated charts tend to be chart shaped rather than argument shaped. They plot what you gave them instead of making the one comparison that wins the point.
The workflow that actually works
- Use the tool to get a first structure out of your head and onto slides.
- Rewrite every headline yourself as a full sentence. This is where the thinking happens.
- Cut a third of the slides. Generated decks are always longer than they need to be.
- Then hand it to a designer, if the deck matters, with the argument already settled.
That last step is cheaper than people expect, because a deck that arrives with a settled argument is a design job rather than a strategy job.
When to skip the tool entirely
If the deck is going to an investor, a board or a big customer, start from the story rather than from a generated draft. Not because the tool is bad, but because a generated draft anchors you to a structure you did not choose, and unpicking that takes longer than writing the thing properly in the first place.