Plan a research talk with AI
Turn a paper into a timed talk with a clear takeaway, speaker notes and backup material. Choose the appropriate slide source or outline for your presentation platform.
To plan a research talk with AI, decide what someone should remember on the train home. A conference slot is rarely long enough to walk through an entire paper. The talk-builder skill helps an agent work from the research question towards a defensible takeaway, with a role and a time allowance for each slide.
The same finding may need two different introductions. A specialist audience can often recognise the method from its name; a public lecture may need a simple example before the first chart. Keep details that answer likely questions in reserve instead of squeezing everything into the spoken narrative.
What it does
- Choose a figure that can carry the main result on its own.
- Write speaker notes in language you can actually say aloud.
- Set one transition that explains why the next finding matters.
- Prepare a backup slide for a predictable methodological question.
- Mark a safe cut in case an earlier explanation takes longer than planned.
Install and use
- Supply an authorised paper, audience description, speaking time and desired slide format. The host can prepare text sources for formats such as Marp or Beamer.
- Ask “Plan a twelve-minute talk around the second result. Put the sensitivity analysis in backup and give me a rehearsal schedule.”
- Deliver a timed practice run. Cut or simplify any section that crowds out the final point.
FAQ
Source
Made by Maya Arazi. Original skill: talk-builder, published on GitHub under the MIT licence.
View on GitHub