Stop retyping customer-specific order rules from SOPs: have AI agents read each client's procedures, recommend the field changes on every new order, and wait for a person to approve before anything updates: AI agent prompt
Thoughtworks published a Zones client story on turning unstructured order SOPs into a production agentic workflow on Microsoft Foundry / Azure OpenAI. Zones CSMs used to manually review every enterprise order against customer-specific procedures (up to 17 touchpoints). The new event-driven flow interprets SOPs into conditions/actions, detects new orders, has AI agents recommend executable field changes with SOP evidence, and stages those recommendations in Azure SQL for CSM approval in a pop-up—never applying changes automatically. Reported outcomes include ~50% faster verification (about one minute down to 10–15 seconds), ~90% recommendation accuracy, 29 enterprise customers on the workflow, and a reusable platform for later agentic ops.
The prompt
1. Pick one order or fulfillment process where staff still copy rules from written SOPs or tribal knowledge into your OMS/CRM on every order. 2. Put each customer's current SOP in a versioned folder (or knowledge base) and write which fields the AI may recommend changing. 3. Connect an AI agent platform you already use (Microsoft Foundry, Bedrock, Claude with tools, etc.) so that on each new order it loads that customer's SOP + the order payload and returns proposed field changes with a quote from the SOP for each. 4. Stage recommendations for human approval: show current value, proposed value and evidence; only write to the system after a person clicks approve. If confidence is low, send the order to the normal manual path. 5. Log every recommendation and decision for a week, measure time-per-order and error rate, then expand to the next customer SOP set.
How to use it
- Copy the prompt above.
- Replace every [bracket] with your own details (or fill in the blanks on AgentPlaybook Real Estate, which remembers your answers for the next prompt).
- Paste it into Microsoft Foundry / Azure OpenAI (or Bedrock/Claude agents), your OMS or Dynamics 365, versioned SOP store and review the answer before you use it.
Works with: Microsoft Foundry / Azure OpenAI (or Bedrock/Claude agents), your OMS or Dynamics 365, versioned SOP store, approval UI or ticket queue
Real-world example: thoughtworks.com/en-us/clients/transforming-unstructured-operational-k
Featured in: AI prompts for writing SOPs, reports, policies and checklists
Before you use it: what to check
Already built into this prompt
- Stage recommendations for human approval: show current value, proposed value and evidence; only write to the system after a person clicks approve.
Check yourself
- Check every figure, price and total against your own records before anything is sent, posted or booked. A person approves refunds, discounts, payments and price changes.
- Remove names, contact details, account numbers and other personal data the task doesn't need before you paste, and check your AI tool's data and training settings.
- Read and edit the draft before it reaches a customer or goes public. If you later automate this prompt, start with "draft and notify me", not "send automatically".
- Check names, dates, numbers and any claim against your own source material before you use the answer. AI can misread or invent details.
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.
Want to make this prompt even better? Grade it with the free prompt grader or build your own with the AI prompt generator.
32 copy-paste agent prompts (PDF) plus 5 new prompts every week.
Get it free