Deck design research — what I found, what changes

Prepared by Cairo

Two tracks: the skills that already exist here, and what the presentation-design field teaches. Both point at the same weakness in draft one.

Skills inventory

What exists, and the verdict on each

SkillVerdict for deck work
pptx (Anthropic)Adopt its design rules — every slide needs a visual element; one color dominates (60–70%); dark/light "sandwich" structure; vary layouts. Its PPTX machinery also becomes our export path if a client ever needs .pptx.
frontend-designAdopt its method — the "signature element" discipline (spend boldness in ONE place), structure-is-information (kill decorative devices), self-critique against the three AI-default looks. This is also your own recorded fix for AI-slop design.
datavizAdopt when charts appear — the chart discipline decks.md already points to; no data slides in this deck yet.
canvas-design (Anthropic)Wrong tool — builds abstract art objects, not claims-disciplined decks. One idea travels: craft density that rewards a second look.
design (Claude Design canvas)Not now — visual click-to-edit surface, but early preview; our artifact deck already gives you+Stacy in-browser copy editing with the claims machinery intact.
Marketplace searchNo deck-specific skills exist beyond these — nothing to install.
Field research

What the presentation-design canon teaches

Iteration 2 — the plan

Same copy, same claims. Five design changes.

Afterwards, the lessons that survived the build get written into Cairo's decks.md — that's the training-run loop closing.

← Roundup