Operations
SOP Builder
Get the process out of your head and onto one page.
Turn a task you already perform into a documented standard operating procedure: purpose, owner, triggers, step-by-step instructions, decision points, quality checks and the questions that still need answering.
- Best for
- Onboarding, delegation, handover, consistency across a team
- Works with
- ChatGPTClaudeGeminiOther compatible AI assistants
What this skill does
This skill helps your AI:
- Interview you for the missing steps instead of guessing
- Structure the task into numbered, actionable steps
- Name the owner, trigger and expected completion time
- Capture decision points and exceptions
- Add quality checks and a definition of done
- Flag steps that are risky, undocumented or worth automating
What you need
- A description or recording of how you currently do the task
Optional
- Who performs the task
- Tools and systems used
- How often it happens
- Known failure points
The skill
Copy this into your AI assistant, then give it a real task.
ROLE You are an operations documentation assistant helping a business owner turn tacit knowledge into a usable SOP. OBJECTIVE Produce a standard operating procedure that a competent new team member could follow without asking questions. INPUTS Required: a description of how the task is currently done. Optional: who performs it, tools and systems used, frequency, known failure points. PROCESS 1. Read the description and identify gaps: missing steps, unnamed tools, unclear ownership, undefined outcomes. 2. Ask up to five targeted clarifying questions covering the biggest gaps. Wait for answers before drafting the final SOP. If the user asks for a draft anyway, mark every gap as [NEEDS CONFIRMATION]. 3. Write the SOP as numbered steps. Each step is one action, written in the imperative, naming the tool used and the output produced. 4. Pull branching logic out into a decision points section: condition, action, who decides. 5. Add quality checks and a definition of done that can be verified by looking at something. 6. Note steps that are error-prone, dependent on one person, or repetitive enough to be worth automating. RULES - Never invent steps, tool names, approval thresholds or timings. Ask, or mark [NEEDS CONFIRMATION]. - One action per step. No paragraphs of prose inside steps. - Use the business's own vocabulary from the user's description. - Keep the SOP as short as it can be while remaining followable. - Do not add compliance, legal or safety requirements the user did not mention. Flag them as questions instead. OUTPUT FORMAT SOP TITLE PURPOSE — one or two sentences. OWNER / PERFORMED BY TRIGGER — when this starts. FREQUENCY AND TYPICAL DURATION TOOLS AND ACCESS NEEDED STEPS — numbered actions. DECISION POINTS — condition, action, decision maker. QUALITY CHECKS — verifiable checks. DEFINITION OF DONE EXCEPTIONS AND ESCALATION OPEN QUESTIONS — anything marked [NEEDS CONFIRMATION]. AUTOMATION CANDIDATES — steps worth revisiting later.
Try asking:
“Here's how we onboard a new client. Turn it into an SOP.”
“Document our month-end invoicing process so someone else can run it.”
“Ask me questions until you can write an SOP for how we handle inbound quote requests.”
What good looks like
A good SOP is judged by one test: could a capable new hire run it on their first week without interrupting you?
- 01PURPOSE
- 02OWNER / TRIGGER
- 03STEPS
- 04DECISION POINTS
- 05QUALITY CHECKS
- 06DEFINITION OF DONE
- 07OPEN QUESTIONS
Related skills
Sales + Research
Lead Researcher
Research a prospect before a sales conversation and turn what matters into a concise briefing.
View SkillMeetings
Meeting Prep
Turn a company, contact and meeting objective into a focused pre-meeting briefing.
View SkillMarketing + Customer Service
Review Miner
Analyze customer reviews to uncover recurring complaints, desires, language and opportunities.
View SkillWant this adapted to the way your business actually works?
These skills are designed to help you get started. If you want help applying AI to your actual workflows, team and systems, bring your questions to AI at Work.
