SOP generation and upkeep turns how a task gets done, from a screen recording, a walkthrough call, or a rough description, into a step-by-step procedure stored where your team looks. It flags when a tool or step changes and asks an owner to confirm the fix. A person approves every version. Most teams are live in two to three weeks.
The problem
The way work actually gets done lives in three places, none of them useful. Some of it is in one person's head, so nothing happens the week they are on holiday. Some is in a Slack message pinned eight months ago. And some is in a doc somebody wrote in 2022, where half the screenshots show a tool that has since been redesigned and two of the steps are just wrong now. New people learn by asking. Everyone else learns by getting it wrong once.
Put a number on how rare current documentation is. In its report The State of Business Process Management 2016, BPTrends (Business Process Trends) found that only 4 percent of organizations said their business processes are documented and kept up to date 100 percent of the time, while half (50 percent) said they do so only occasionally, meaning 1 to 30 percent of the time. Scale the cost to a firm size: across a 40-person company, we typically model 2 to 3 hours a week per person lost to asking how something is done, re-explaining it, or redoing work off an outdated version. That is 80 to 120 hours a week, most of it invisible. (The 4 percent and 50 percent figures are from BPTrends. The 2 to 3 hours per person is a modeled estimate; 2 to 3 times 40 people is the 80 to 120 hours.)
The part that never shows up on a timesheet is the wrong version being followed. A billing step changed in the tool but not in the doc, so an invoice goes out on the old process. A new hire takes two months to get up to speed because there is nothing to read, only people to interrupt. Work gets redone because the first version followed a step that stopped being true. A procedure nobody trusts is quietly worse than no procedure, because at least with no procedure people ask.
How the automation works
It captures how the work is really done.
You hand it a screen recording of the task, a transcript of a walkthrough call, or a rough written description, and it drafts a clean step-by-step procedure from what it saw, in plain language, with the steps in order.
It formats and stores it consistently.
Every SOP comes out in the same shape, with the same headings and structure, and lands where your team already looks, in Notion, Google Docs, or your wiki, linked to the tools and steps it references.
It watches for drift and prompts an owner.
When a referenced tool changes or a step stops matching how the work is done, it flags that SOP as possibly stale and asks the named owner to confirm the update. Nothing changes until a person approves it.
The pieces are proven: drafting a procedure from a recording or transcript, formatting to a consistent template, storing in your wiki, and sending a notification when something looks out of date. The real work is the wiring. An SOP that quietly goes stale is worse than none, because people follow the wrong steps with full confidence, so the value is not the first draft, it is the upkeep and the human owner behind it. The hard part is detecting when reality has drifted from the document, and getting a busy owner to actually confirm the fix instead of ignoring the flag. That detection logic, the ownership model, and the review step are what get set up, tuned to your processes, and handed over during implementation.
What this looks like in practice
Client onboarding and billing procedures live in two senior people's heads, plus a stale doc nobody opens.
- A new operations hire takes about 8 weeks of shadowing before they can run client onboarding without checking in on every step.
- A billing step changed in the tool three months ago and the doc still shows the old screen, so two invoices went out on the wrong process before anyone caught it.
- The operations lead loses about 4 hours a week answering "how do we do this again" and rewriting bits of the doc after the fact.
- Procedures are drafted from screen recordings and one walkthrough call, formatted the same way, and stored in Notion where the team already looks.
- When the billing tool's screen changed, the SOP was flagged as stale and the owner confirmed the corrected step in minutes, before it produced another wrong invoice.
- The next new hire reaches independent onboarding in about 4 weeks, and how-to interruptions drop off.
Typical impact
Typical ranges for this pattern, not client claims. Your numbers get modeled in the audit.
Systems it connects
Plus most tools with an API. The audit maps your exact stack.
Who this fits
- Repeatable procedures that live in a few people's heads, or in docs no one trusts
- 10 or more employees, with onboarding or handoffs that depend on "go ask Sarah"
- Real processes worth capturing: onboarding, billing, fulfillment, support, compliance steps
- Someone to own each SOP, because upkeep needs a person to confirm changes when a step drifts