Every quarter we promised ourselves we would finally document our processes. Every quarter, something more on fire took priority, and the documentation promise quietly migrated to next quarter's list of promises. By the time we crossed thirty employees the tribal-knowledge risk had stopped being a theoretical bullet on a risk register: three people held the entire mental model for client onboarding, and two of them had started casually dropping the phrase 'when I eventually move on' into meetings. That is the corporate equivalent of a smoke alarm chirping at 3 a.m. — easy to ignore for exactly as long as it takes for everything to catch fire.
The real bottleneck was never knowing what to write
We always knew which processes needed documenting. That part was obvious. The actual blocker was the sheer grinding labour of writing a good SOP from scratch. A real one has a scope, numbered steps, named roles, decision branches, and quality checks. Done properly, that is forty-five to ninety minutes of focused, uninterrupted writing per process — and 'focused and uninterrupted' is not a state that exists naturally in a thirty-person company. Multiply ninety minutes by thirty processes and you have a six-week project that no manager on earth will ever fund, because 'write internal documentation' loses every single prioritisation meeting to literally anything with a revenue number attached to it.
The afternoon that cleared two years of debt
I booked a room, ordered an irresponsible quantity of coffee, and gathered our four department leads. I gave each one twenty minutes to brain-dump the rough flow of their top processes — the exact kind of thing they would scribble on a whiteboard while explaining it to a new hire who is nodding too fast. Then I fed each rough description into the SOP Generator from a Rough Process Description prompt running on Claude Opus 4.8. Opus is genuinely the right tool here for two reasons: it follows the XML-style structure tags without drifting off into improvisation, and it reasons carefully about the implied decision branches that the lead forgot to mention out loud but clearly exists in how they actually work.
Every brain-dump came back as a fully structured SOP — purpose, scope, definitions, a roles table, numbered steps with real decision branches, quality checks, and a common-errors section that read like it had been written by someone who had watched the process fail before. The first one genuinely took my breath away a little. By the fourteenth, I had stopped being impressed and started being briskly efficient, which is the sign of a tool that works: it stops feeling like a demo and starts feeling like a Tuesday.
- Average generation time was under thirty seconds per SOP, including the careful reasoning step
- Human review and cleanup ran ten to fifteen minutes each — mostly fixing tool names and tightening a couple of role labels
- Total time from zero SOPs to fourteen reviewed and published in our internal wiki: about four hours, coffee included
- Not one of the fourteen needed a second full draft; every edit was surgical rather than a rewrite
The homework moment
Here is my favourite part of the whole afternoon. The prompt has a rule baked in: if the description is too thin to produce a genuinely useful SOP, it returns three clarifying questions instead of faking a document. Three of my four leads did not get an SOP back on at least one of their processes. They got a short list of pointed questions about their own workflow — and discovered, live, in the room, in front of their peers, that they could not actually answer them. The model flatly refused to let them hide behind a vague description. One lead muttered 'huh, I guess we never actually decided that' about a step he had been doing inconsistently for over a year. That single unanswered question was worth the entire afternoon and most of the coffee budget.
Why this is the one I would pay for
I have a lot of free prompts I love and use daily, but this is the one I would buy without a second thought. The 'never invent a tool name' rule means the output is trustworthy enough to publish without a paranoid, line-by-line, second-guessing audit, which is most of the cost of documentation. And the clarifying-questions behaviour turns documentation from a typing exercise into a thinking exercise — it makes your experts confront the parts of their own process they have been quietly improvising. We onboarded two new hires last month directly off these documents, with effectively zero hand-holding, and that is the number that actually matters.
Takeaway
Process debt is not a discipline problem, it is a friction problem, and this prompt removes the friction so completely that the discipline becomes the easy path. When I need to brief leadership on a rollout that came out of one of these SOPs, I run the summary through my Decision Doc: Options, Risk, and a Clear Recommendation prompt to make the call cleanly. Grab the SOP Generator from a Rough Process Description prompt on Prompt Dock, gather your leads, buy too much coffee, and clear your entire documentation backlog before lunch. The smoke alarm finally stopped chirping.