Turn a project update into a presentation with a point

Build a decision-focused presentation from approved evidence, including a worked slide sequence, claim checks and a rehearsal prompt.

A useful task, worked through.

A presentation needs a reason for the audience to listen. Before asking AI to make slides, state what the audience should understand or decide at the end. Then give each slide one job. A deck that looks finished but never makes the decision clear has only solved the formatting problem.

Write the decision in one sentence

Our fictional team wants approval for a two-week trial of a shared intake form. The current evidence is three examples of missing information in email requests and a draft form. There are no measured savings and no customer results. The decision is whether to run a small trial with a review date, not whether to replace every team process immediately.

Give the assistant the audience’s existing knowledge and the meeting length. A manager deciding on a trial needs scope, effort and success criteria. A colleague implementing it needs field definitions and handling instructions. Combining those audiences can produce slides that answer neither person’s main question.

Build the argument before the slides

An authored five-slide sequence is: the decision requested; the recurring intake problem with the three fictional examples; the proposed form and trial boundary; what the team will observe; and the approval needed to begin. Each slide has a purpose in the argument. The order makes the final request feel earned.

For the observation slide, suggest measures such as missing required fields and requests needing clarification, with a clear counting method. Do not promise a percentage improvement before running the trial. A good presentation can propose how to find out whether an idea helps without pretending the answer is already known.

A request you can adapt

Create a five-slide outline for this decision and evidence packet. Give each slide a claim, supporting evidence and speaker note. Label proposals and fictional examples. Keep missing evidence visible. End with the exact decision and trial boundary. Do not invent charts, testimonials or measured benefits.

Use a visual only when it explains something

A simple before-and-after flow can show an email request becoming a form submission and a review queue. Label the after state “proposed.” A decorative dashboard with fabricated metrics would imply an operating system and results that do not exist. Ask for a visual specification that a reviewer can compare with the evidence.

Inspect the exported deck at the size the audience will see. Check that labels remain readable, charts include units and citations are usable. Open the actual presentation file, not only a preview image. Speaker notes should explain the slide rather than hiding essential qualifications from the audience.

Rehearse the hardest reasonable question

Ask the assistant to play a skeptical colleague who wants to know whether the form will create more work. Answer using the trial boundary and observation plan. If your answer depends on a benefit you have not measured, revise the slide so the uncertainty is explicit.

For practice, reduce the meeting to three minutes. Keep the decision, strongest example, trial scope and approval request. Move implementation detail to an appendix instead of shrinking every font. After rehearsal, have someone unfamiliar with the project tell you what decision you asked for. Their answer is a useful check on whether the deck communicates its purpose.

Build and inspect the finished presentation

Take it further with a changed example

If a deck feels like a list of topics, write the decision it should support at the top of a separate page. For each slide, complete “the reader needs this because…” Remove or relocate material that has no clear role in that decision.

Try reading only the slide titles in order. They should form a coherent argument, with uncertainty still visible. “A limited trial would answer the capacity question” communicates more than “Next steps,” but only if the evidence actually supports that trial.

For a further exercise, cut one slide without changing the decision request. Decide what information must move and what can become an appendix. Then ask a fresh reader to state the requested decision and the strongest unresolved concern from the shortened version.

Take this into your next task

Write the decision and evidence sequence first; give the slides a clear argument to carry.

Build your learning roadmap.

Questions people ask

Can AI make a chart from a qualitative example?

It can suggest a diagram, but numeric charts need real data and clear definitions.

How do I show an untested idea?

Label it as proposed and explain the trial and evidence you would collect.

What should I inspect after export?

Readability, slide order, missing elements, working links and whether the actual file opens correctly.

Should I let AI choose the visuals?

It can propose options, but select a visual that helps explain the specific evidence or process. Inspect labels, numbers and implied claims in the finished slide.

Sources and editorial notes

Reviewed 2026-09-07. Product capabilities depend on the app, plan, region and workspace settings. Workflows and prompts are authored teaching material, not recorded model results or measured time-saving claims.

Keep going

Related Power of AI pages

Keep reading with Start here, everyday uses, the tool guide, the writing workshop, and sources and standards.