GaneevEcoWhy this works better than generic training
Generic prompting tips are hard to apply to a specific job — knowing you should "be specific" doesn't tell a sales rep exactly what a good account-research prompt looks like for their particular workflow. A library of ten to fifteen prompts that already work for exactly what a sales rep, an analyst, or a support agent does every day removes the hardest part — figuring out what to ask in the first place — and lets people learn by editing something that already works, rather than starting from a blank page.
This approach also scales training effort far better than individual coaching does. One well-built library serves every new hire in that role indefinitely, whereas one-on-one coaching has to be repeated for every single new person who joins the team, making the library approach a genuinely better return on the time invested in building it well the first time.
Building one that doesn't go stale
- Start from what your best users are already doing well, not a theoretical ideal that sounds good in a training deck but nobody actually follows in practice.
- Keep it in a shared, owned location — a Project's knowledge base is a natural fit — not a document that quietly goes stale in a forgotten folder nobody revisits.
- Review and prune quarterly; a library nobody curates becomes as unreliable as no library at all, and worse, it erodes trust in the whole approach once people notice it's stopped being useful.
It's also worth designating a specific person per role — not necessarily the most senior person, but often whoever's shown the most genuine enthusiasm for getting good results — as the informal curator who fields suggestions and keeps the library current. That light ownership model tends to work better than a purely top-down, centrally managed library, since the curator is close enough to daily work to know which prompts have quietly stopped being useful and which new ones deserve to be added, long before that information would otherwise reach whoever's nominally in charge of the rollout from a greater organizational distance.
It's worth building a lightweight process for retiring prompts too, not just adding new ones. A library that only ever grows, without anything ever being removed, becomes harder to navigate over time and eventually undermines the very simplicity that made it useful in the first place. Treating removal as an equally legitimate curation action, alongside addition, keeps a library genuinely useful years into a rollout rather than becoming a sprawling, unmaintained archive nobody has full confidence in anymore.
A simple quarterly rule — anything not used or referenced in the last quarter gets flagged for review — keeps this manageable without turning curation into its own significant ongoing project.