How to Design Charts in a Presentation With AI

The input is always your data. That might be a table you wrote directly in your content, rows pasted from a spreadsheet, or numbers you type out in a sentence, "revenue was 2.1 million in Q1, 2.6 million in Q2, 3.4 million in Q3." Eazy reads uploaded Excel and CSV files too, so a chart can come straight from a spreadsheet you already have without retyping the numbers by hand.

From there, you describe the comparison, not the pixels. "Turn this into a bar chart," "chart the trend over the four quarters," "show these five categories as a share of the whole" all give the tool enough to build the right shape. The chart it returns is a real rendered chart, drawn from your actual numbers, not an AI-generated picture that only looks like a chart.

Because the chart lives inside your document like any other content, changing the underlying numbers is a content edit, not a redesign. Update a row in the source table or tell Eazy the new figure, and the chart that depends on it rebuilds, while the rest of the slide and the rest of the deck stay exactly as they were.

1. Bar chart, for comparing categories side by side. Use it when the point is "this is bigger than that": revenue by region, signups by channel, scores by team. A bar chart makes the size difference visible at a glance, which is exactly the job a table of numbers forces the audience to do in their head.

2. Line chart, for a trend over time. Use it when the story is direction, growth, decline, or a turning point across weeks, months, or quarters. A line chart makes the shape of the change the point, not any single number on it.

3. Pie or donut chart, for share of a whole. Use it only when the pieces genuinely sum to one total and there are a handful of slices, five or fewer reads cleanly, ten does not. Beyond that, a bar chart of the same categories almost always communicates faster.

4. Area chart, for cumulative or stacked trends. Use it when you are showing a trend and also how it breaks down into parts, total revenue over time with each product line stacked inside it, for instance. It carries more information than a line chart at the cost of being slightly busier.

5. Scatter chart, for the relationship between two variables. Use it when the point is correlation or clustering, price against satisfaction, spend against outcome, not a simple ranking or trend. It is the least common choice, and the most useful one when the question is genuinely "do these two things move together?"

In every case, name the comparison you are making when you ask for the chart, categories side by side, trend over time, share of a whole, relationship between two things, rather than just "add a chart." The specific ask is what gets you the type built for your point instead of a default applied to whatever data was nearby.

One chart, one point. A slide that tries to show revenue, headcount, and churn all at once forces the audience to decide what matters. Split it: one chart per slide, each doing a single job, reads faster than one crowded slide trying to do three.

State the takeaway, do not just show the data. A chart shows what happened; a headline or callout says why it matters, "churn dropped 18% after the new onboarding flow," sitting above or beside the chart. Because slides are designed on-brand by default, that headline and the chart are placed to work together, not as two separate elements competing for attention.

Let the chart inherit your theme. Apply a theme to the deck and chart colors, type, and spacing restyle along with everything else, so a chart does not look like it was pasted in from a different tool. This is where AI charts genuinely beat a spreadsheet screenshot, the chart is a native part of the design system, not an image dropped on top of it.

Watch for the classic data-slide mistakes regardless of the tool: too many categories on one axis, a legend doing work a direct label could do faster, or a chart type chosen because it looked interesting rather than because it matched the comparison. A simple bar chart that lands the point beats a fancier chart that makes the audience work to read it.

This distinction matters because it changes what you can promise in a meeting. If you are sharing a link and presenting from the browser, or letting someone open the deck on their own, hover tooltips are genuinely there, real interactivity, not a marketing description of it. Someone can hover a bar and see the exact figure without you narrating every number.

The moment you export, that behavior goes away. PDF and PPTX are static formats by nature, so the chart bakes down to a clean image at export time, visually the same chart, just no longer hoverable. This is not a limitation unique to Eazy; it is true of any live web chart moved into a static file format, and it is worth knowing before you promise a client an "interactive PDF."

The practical rule: if the room has a screen and a live link, present from the browser and let the interactivity do some of the talking. If you need a file to email or print, export knowing the chart becomes a picture, still styled, still on-brand, just fixed rather than explorable. Plan which version you need before the meeting, not during it.

1. Bring in your data. Drop an Excel file, a CSV, or paste a table straight into your content, or type the numbers directly if there are only a few. Eazy reads it into editable content either way, so the numbers are not locked inside a picture of a spreadsheet.

2. Say what you are comparing. Instead of "make a chart," say what the chart needs to show, categories side by side, a trend across time, share of a total. That single sentence is usually enough for the right chart type to come back.

3. Let it design the chart into the slide. The chart that comes back is placed and styled as part of the slide, not dropped in as a separate image, on-brand by default the same way the rest of your deck is.

4. Refine by chat, one slide at a time. If the first chart is not quite right, say so in plain language, "swap this to a line chart," "drop the smallest three categories," "make the headline state the takeaway." Only that slide rebuilds, so the rest of the deck you already approved is untouched while you dial in the one chart that needs work.