Status narratives, risk logs, scope memos, stakeholder updates, closure reports. Every prompt carries a note on what good output looks like and what the model reliably gets wrong — because a prompt without acceptance criteria is just a faster way to send something you haven't checked.
One-off payment · no subscription · 12 months of updates · 30-day refund
"Write me a project status report" gets you eighty words that would be true of any project in any week. Your sponsor reads it, learns nothing, and quietly stops opening your updates.
The prompts here are engineered to force specific behaviour out of the model — refusing to invent figures, marking what it couldn't determine, separating what you told it from what it assumed. That's the difference between a draft you can send and a draft you have to rewrite.
Workday surveyed 3,200 people: AI saves most of them one to seven hours a week, and roughly 37% of that is lost correcting and rewriting the output. Four hours lost for every ten gained. Almost all of it comes from prompts that were never designed to be checkable.
Not by technique. You arrive with a status report due Thursday, and you find the status report prompts.
The honest status narrative. The RAG justification test. Three registers, one message. The missing-information audit.
Risk log triage. Pre-mortem. Assumption extraction. Escalation drafting.
Notes to RAID. The decision record. Meeting purpose test. Action chase, written well.
The scope-change memo. Creep detection. Requirement ambiguity scan. Acceptance criteria.
The bad-news message. Stakeholder position map. The difficult conversation script. Technical to commercial.
Estimate interrogation. Work breakdown from a brief. Critical path in plain English. The replanning options paper.
The workslop detector. Document consistency check. Reverse brief.
Lessons that are actually lessons. The handover pack. The closure narrative.
On every single prompt. How to judge what came back in under thirty seconds, so you know whether to send it or run it again.
Provenance, substance, judgement, accountability. Four minutes per artefact, and it's where the rework tax actually gets recovered.
A status narrative, a risk log triage and a scope memo — before and after, including the one where the model invented a mitigation.
Everything that helps you decide whether AI is safe to use costs nothing. You pay for the part that saves you time.
Because 78% of people use AI without approval and only 37% of organisations have any policy — and that's a problem you shouldn't have to buy a product to fix. The classification card and the policy cost us nothing to give away and they're genuinely useful on their own. Take them and never come back if that's all you need.
Because you'd be paying monthly for something you use once. You'll copy the six prompts that match your artefacts into your own document in the first week, and then a subscription is just a charge you forget to cancel.
So it's one payment. You keep it. Updates come free for a year. If a version arrives later that genuinely earns a recurring price — a shared team workspace, your own prompts stored alongside ours — that'll be a separate thing you can choose, not a default you're enrolled in.
If the library saves you one hour a month — a single status narrative you don't rewrite — it has paid for itself before the end of the first quarter. That's a low bar, deliberately. We'd rather set one you'll clear than a number you'd have to take on faith.
Excluding VAT, which is added at checkout based on where you are.
Thirty days, full refund, no form and no questions. If you read three prompts and think you could have written them yourself, say so and we'll refund you — we'd rather have that feedback than the €49.
Whichever your organisation has approved. Everything works with ChatGPT, Claude, Copilot, Gemini or a local model. If nothing is approved yet, start with the free policy card — that situation is exactly what it's for.
Yes, and some of you will. What you're buying is the version that has already been run on real project material and revised — plus the acceptance note and the failure modes, which are the parts nobody writes for themselves because you only learn them by getting it wrong on something that mattered.
Project. Delivery, PMO, agency and consulting work. Product management is well served already; non-tech project management is not, and that gap is why this exists.
Parts of it. That's why the prompts are built around artefacts rather than tools, and why updates are included for twelve months. A status report will still be a status report; whichever model you're using won't be the same one.
One licence is one person. If you want it for a team, email and we'll sort something out — there's no team tier built yet and we'd rather agree a fair price than pretend there's a page for it.
Updates for a year, and more prompts as they're written. If a team workspace gets built, existing buyers get told first and won't be auto-enrolled in anything.
Twelve minutes, twenty questions, a score out of 60 and the one thing worth fixing first. If the library isn't what you need, the assessment will tell you that too.
Take the assessment Free · no signup · nothing leaves your browser