Paper presentation slides:decide each slide's job before pasting the AI summary

Start by getting clear on what a slide is for.

A slide is neither your script nor your handout. It's a tool for directing the audience's eyes to the one thing you want them to look at right now.

Don't make people read — show them. Explain with your voice, and let the slides play a supporting role. That division of labor frees you from the pressure of cramming everything onto the screen.

There is only one design principle you need to remember: each slide should do exactly one job.

"This slide conveys the study design." "This slide shows the main result." When every slide has a single defined purpose, the audience never gets lost.

Conversely, when methods, results, and discussion all share one slide, the viewer's eyes wander and nothing sticks.

The practical trick is to list only each slide's job — like sticky notes — before you open your slide software at all. Once the jobs are laid out, filling in the content is the easy part.

A paper presentation can work with as few as six slides.

Slide one: why you chose this paper — your selection reasoning and the clinical or research question behind it. Slide two: what the study investigated, including the PICO and study design.

Slide three: how the study was conducted — the essential points of the methods. Slide four: the figures and tables that carry the logic of the results.

Slide five: how much to trust it — the limitations. Slide six: the take-home message, closed with a single sentence.

For short presentations, doing each job well beats adding more slides. This minimal build works well for brief paper introductions, as distinct from the fuller deck you'd bring to an in-depth journal club.

The figures slide is the heart of the presentation. Here are three things to avoid.

First: using a figure without checking the original. Before you crop or highlight anything, check the axes, units, group labels, and annotations in the original PDF. If you edit a figure while misreading its meaning, the error gets amplified on screen.

Second: tiling several small figures side by side. Put two or three figures on one slide and none of them will be readable. Don't drop figures the logic needs — but give each one its own slide, shown large.

Third: copying numbers from the AI summary. Always take effect sizes and p-values from the original text and figures before they go on a slide. AI can round or misread numbers.

A common mistake is tacking "Limitations" in small print at the bottom of the results slide.

That's a waste. When results and limitations share a screen, the audience gets only a shallow grasp of both.

Limitations aren't a footnote to the results — they are independent information that determines how far, and for whom, those results can be used.

Let the results slide name the primary outcome clearly and show the numbers with confidence. Then let the limitations slide narrow down where those results apply. This two-step structure is both honest and easy to follow.

AI summaries are an excellent starting draft for slides. But you should edit them, not paste them.

Summary sentences are written in an explanatory register — "the study showed that...". You need to compress them by hand into bullet points and short phrases that belong on a slide.

For example, rephrase a full sentence as "More bleeding events in the elderly group," adding the numbers you've verified in the original where needed.

This compression step feels tedious, but it's actually the core of preparation. Anywhere you can't rephrase in your own words is a place your understanding is thin — a signal to go back to the original.

If you use Paperfy, slide preparation can look like this.

When you upload a PDF, the AI summary and the extracted figures appear on the same paper page. That makes it easy to compare the order and relationships of the figures, and once you've chosen the ones you need, you can jump back to the corresponding page of the original PDF on the same screen to verify axes and labels.

Then, if you edit the AI summary by hand into the compressed phrasing you built for your slides, that slide-ready summary stays in your library after the presentation. The next time you cite the same paper, you won't be starting from scratch.

Don't make more slides — design each one around a job.

Give every slide a single purpose. Verify the axes, units, and labels of your key figure in the original. Take numbers from the original text, not the AI summary. Put results and limitations on separate slides.

For your next presentation, start by laying out six sticky notes — one job each.