27: The AI Library - Prompts and Skills - transcript
Liam: The QUILL Cast, episode twenty-seven. I'm Liam. Today the AI stops being a set of buttons and becomes a workshop: the AI Library, prompts, and skills, taught as a build-along.
Jessica: I'm Jessica. The Library, under the AI menu, is one tabbed home for three things: Prompts, Skills, and Agents. Today covers the first two rungs; agents get their own episode next.
Liam: Our running example, borrowed from a real listener: every week, rough project notes must become a status update for a manager. Same shape weekly, different content. Watch it climb the ladder.
Jessica: Rung one: a Prompt, a reusable template with variables. New prompt in the Library: rewrite the following notes as a professional status update with three sections, Progress, Blockers, Next Steps, under two hundred words. Add the input variable for the text. Save.
Liam: Now it's a menu item. Monday: select messy notes, run the prompt, review, accept. Thirty seconds, forever. That's the whole rung-one skill: noticing a transformation you type repeatedly into Ask Quill, and saving it with a hole for the input.
Jessica: Prompts travel, too: they export as small P Q P pack files. Build the status prompt once, share it with the team, and everyone's Monday shrinks. Community packs exist for exactly this reason, same culture as episode nine's keyboard packs.
Liam: Rung two: Skills, multi-step workflows where each step feeds the next. Our example grows up: step one, summarize the notes into factual bullets. Step two, rewrite the bullets in house style. Step three, generate a subject line and prepend it.
Jessica: Why steps instead of one mega-prompt? Control and reliability. Models, especially the free-tier ones from episode twenty-five, do dramatically better on three small jobs than one tangled one. And when output disappoints, you can see which step went wrong, instead of shaking a black box.
Liam: The delightful mechanical detail: there's a real Promote button. Your prompt graduates into step one of a new skill, the ladder is built into the tooling, you never start over. Skills export too, S Q P packs.
Jessica: When to climb: one transformation, stay a prompt. Distinct stages, or structure you want enforced identically every run, make it a skill. And if the steps themselves would depend on what's in the document, that's rung three, next episode.
Liam: Worth studying: the built-in prompt library that ships with QUILL. Reading how the provided prompts are phrased, specific, structured, explicit about output shape, is a free course in getting good results. Copy their style before inventing your own.
Jessica: And connect the dots backward: episode twenty-three's custom Transcript Actions? Prompts wearing a domain costume. The Action Builder writes what the Library manages. You've been building rung one already; today it just got a name.
Liam: The mindset shift this episode is really about: from using AI features to owning AI tools. A prompt library that fits your actual week, five, six entries, deeply yours, beats a hundred generic ones. Curate like a craftsperson: name things well, prune what you stopped using, share what proved out.
Jessica: Homework, the build-along. One: create the status-update prompt, or your own weekly equivalent, and run it on real input. Two: promote it into a two-step skill, summarize then style, and compare output quality against the single prompt, genuinely compare, run both. Three: read three built-in prompts and steal one phrasing trick. Four: export your first pack, so sharing is a habit from day one.
Liam: Next episode: rung three, agents, plans you review before they run, and the accessibility tune-up that shows the whole idea at its best.
Jessica: I'm Jessica.
Liam: I'm Liam. Build your bench.