Before you hand your service business to AI agents, write down how you actually do the work: turn your know-how into step-by-step workflow docs each agent follows, and review every output at first: AI agent prompt

Agents can only run a business the way you do if your know-how is written down. Emergent reported (Oct 23, 2026) that Sydney recruiter Alicia Keenan rebuilt her firm around four AI agents that find new business, produce market intelligence, nurture candidates and run searches, after spending months writing about 200 workflow documents to capture nearly two decades of expertise. The agents still make mistakes (garbled early reports, candidates shortlisted from the wrong country, emails in the wrong tone), so she checks every interaction and fixes the workflow each time something goes wrong, and expects about 18 months before she can step back from close supervision. This prompt helps you write the first workflow docs and the checks around them.

Curated by AgentPlaybook Real Estate · How we make and check prompts

HR & PeopleRecruitingadvanced

The prompt

You are helping me turn my business know-how into workflow documents an AI agent can follow. My business: [e.g. specialist recruitment for finance roles in Sydney]. The jobs I want agents to take on: [e.g. finding new clients, market updates, keeping candidates warm, running a search]. Start with this one job: [name it]. Interview me one question at a time until you understand: the trigger that starts it, every step I take, the tools and data I use, how I decide (with real examples of good and bad cases), the hard rules (e.g. only candidates with the right to work in [country]), the tone of anything sent under my name, and what 'done well' looks like. Then write the workflow doc with: purpose, inputs, numbered steps, decision rules, things the agent must never do, a draft-for-approval point before anything goes to a client or candidate, and a checklist I use to review each output. Finish with a log template where I record each mistake and the rule we add to stop it happening again. Then ask which job to document next.
Open in ChatGPT Open in Claude Perplexity Download as agent skill (SKILL.md)

How to use it

  1. Copy the prompt above.
  2. Replace every [bracket] with your own details (or fill in the blanks on AgentPlaybook Real Estate, which remembers your answers for the next prompt).
  3. Paste it into Claude, ChatGPT, Notion and review the answer before you use it.

Works with: Claude, ChatGPT, Notion, Google Docs, Bullhorn, LinkedIn Recruiter

Real-world example: emergent.sh/news/ai-agents-enable-one-person-companies-real

Featured in: ChatGPT prompts for business email, replies and follow-ups · AI prompts for writing SOPs, reports, policies and checklists

Before you use it: what to check

Check yourself

Using an AI agent? Download this prompt as a SKILL.md agent skill with these checks built in, or get the .zip for Claude.ai. How to install skills.

Share:Post on XLinkedInFacebookRedditWhatsApp
</> Embed this prompt on your site

Paste this HTML into your blog or website. It's a plain card with no tracking scripts that links back to this prompt.

Want to make this prompt even better? Grade it with the free prompt grader or build your own with the AI prompt generator.

Get the free Solo Founder Stack pack

32 copy-paste agent prompts (PDF) plus 5 new prompts every week.

Get it free

More HR & People prompts

Screen candidates with a voice agent firstRecruiting · intermediateNew hire onboarding planOnboarding · beginner · worked exampleJob description + screening questionsHiring · beginner · worked exampleEmployee feedback analyzerFeedback · intermediatePre-payroll missing data checkPayroll · beginnerOpen roles at-risk watchRecruiting pipeline · intermediate

See all HR & People prompts →

Related prompts from other teams

Draft replies from your FAQCustomer Support · beginner · worked exampleCall notes to follow-up emailSales · beginner · worked exampleLead enrichment + draft replySales · intermediateWeekly newsletter drafterContent · beginner · worked example