Iro AIWhat you'll be able to do
- Write a prompt that produces a deck you can actually present
- Decide between presentation, document, and webpage mode
- Avoid burning free credits on drafts
- Export in the format that survives the handoff
Inside the path
A focused set of five-minute lessons. Each one ends with a hands-on exercise, not a quiz you can guess.
Prompt the structure, not the topic 5 min
Why "a deck about our Q3 results" produces filler and what to write instead.
Cards, not slides 5 min
What changes when the container grows to fit the content.
Credits are finite 5 min
Where the 400 free credits go and how to not waste them.
Exporting cleanly 5 min
PDF, PPTX, and what breaks in each.
Try a sample exercise
This is the kind of card you'd practice inside Iro: you do the thinking, then get feedback.
◆ Sample exercise · Prompt practice
You need an investor update deck covering last quarter: revenue up 40%, two enterprise customers signed, churn slightly worse, and a hiring ask.
Your task: Choose the prompt most likely to produce a deck you can present without rebuilding it.
- "Create a professional investor update presentation for our startup's Q3 results."
- "8-card investor update. Card 1: headline, revenue up 40% QoQ. Cards 2 to 3: the two enterprise logos signed and what they unlock. Card 4: churn rose from 3% to 4%, with the cause and the fix. Cards 5 to 6: hiring ask, two engineers, with the reasoning. Card 7: risks. Card 8: what we need from you. Direct tone, numbers in headlines, no filler."
- "Make the best possible investor deck. Use your judgment on structure and make it visually stunning."
- "Q3 results: revenue, customers, churn, hiring."
See why the second prompt wins
The second prompt wins because it supplies the two things Gamma cannot infer: the structure (how many cards and what goes on each) and the specifics (40%, 3% to 4%, two engineers). Given only a topic, the model fills space with generic business language, which is where the "AI deck" look comes from. Notice it also carries the uncomfortable number. A deck that hides worsening churn does not survive the first question, and telling the tool to include it is how you keep the argument honest. The vague prompts produce something that looks finished and says nothing.
The prompt decides everything
Gamma's output quality tracks the specificity of your prompt almost linearly. Give it a topic and you get filler with nice typography. Give it a structure and real numbers and you get a draft worth editing.
The pattern that works: state the number of cards, say what belongs on each, include your actual figures, and set a tone. It feels like more work than typing a sentence, and it is, but it replaces the twenty minutes you would otherwise spend rewriting generic content.
Cards are not slides
Gamma uses cards that grow to fit their content rather than fixed-size slides. Text does not get squashed to fit a template, which removes the most common cause of ugly decks.
There is also a document mode that produces long-form scrollable content instead of a card-by-card deck, which suits memos, reports, and proposals. Choosing the wrong mode is a common early mistake: if the output is meant to be read rather than presented, document mode is the better container.
Credits, and not wasting them
The free plan comes with 400 one-time AI credits. A full deck generation costs roughly 40, so you get about 8 to 10 generated decks. Free credits do not refresh, though referrals add more, and paid plans provide a refreshing monthly allowance.
The practical consequence: do not use generation as a thinking tool. Work out your structure first, then generate once against a detailed prompt, then edit by hand. Regenerating repeatedly to "see what it comes up with" is how people burn 400 credits in an afternoon and end up with a deck no better than their second attempt.
Exporting without surprises
PDF, PPTX, and PNG export are available across plans. They are not equally reliable.
Because Gamma's native format is a flexible web layout, PPTX export often shifts layouts and substitutes fonts, and complex cards can arrive in PowerPoint as flattened, non-editable images. PDF preserves the design most faithfully and is the safer choice for sharing and printing.
If someone downstream genuinely needs to edit in PowerPoint, plan for that from the start and keep cards simple, or accept that you will be doing cleanup on the other side.