If you've used customer support software before, you've probably encountered macros, canned responses, and rules. They work -- to a point. But they break down the moment a customer phrases something unexpectedly, asks two things in one message, or adds context that doesn't fit the template. AI playbooks are a fundamentally different approach.
What Is an AI Playbook?
A playbook is a set of natural-language instructions that tells the AI how to handle a specific type of customer request. Unlike a macro (which is a pre-written response) or a rule (which is an if-then condition), a playbook is a strategy. It describes the goal, the steps, the edge cases, and the boundaries.
For example, a "Refund Request" playbook might say:
- Check if the order exists and is within the 30-day return window.
- If the customer is asking for a partial refund, calculate the amount for the specific items.
- If the order was placed more than 30 days ago, politely explain the policy and offer a store credit alternative.
- Never process a refund above $200 without human approval.
- Always confirm the refund amount with the customer before executing.
The AI reads these instructions and applies them to each unique conversation. It doesn't match keywords -- it understands what the customer actually wants and follows the playbook's logic.
How Playbooks Differ from Macros and Rules
Macros are static text. "Your order has been shipped and should arrive in 3-5 business days." Useful, but they can't adapt. If the customer asks a follow-up question, you need a different macro.
Rules are conditional logic. "If subject contains 'refund', assign to refund queue." They route tickets, but they don't solve them.
Playbooks are instructions for an AI agent. They describe how to think about a problem, what data to look up, what actions to take, and when to stop and ask a human. The AI generates a unique response every time, following the playbook's strategy while adapting to the specifics of each conversation.
The Role of Intent Detection
Playbooks don't activate randomly. The AI first classifies the customer's message into an intent -- a category like "question," "refund," "cancellation," "complaint," or "WISMO" (where is my order). Each intent maps to one or more playbooks.
When a customer writes "Hey, I ordered a jacket last week and it hasn't arrived yet. Also, can I return the shoes from my previous order?" -- the AI detects two intents: WISMO for the jacket and a return request for the shoes. It activates both playbooks and handles each part of the message separately, in the same response.
Copilot vs. Autopilot
Every playbook operates in one of two modes:
Copilot mode means the AI drafts a response and prepares the action, but a human reviews and approves before anything is sent or executed. This is the default for sensitive operations like refunds and cancellations. You see exactly what the AI wants to do and can edit or reject it.
Autopilot mode means the AI handles the entire interaction independently -- reading the message, looking up data, generating a response, and sending it. This works well for low-risk, high-volume queries like FAQ answers and order status lookups.
The power is in the mix. Most merchants run FAQ and WISMO on autopilot while keeping refunds and complaints on copilot. You can switch any playbook between modes at any time.
Playbook Examples in Action
WISMO Playbook
Customer writes: "Where is my order #1234?" The AI looks up order #1234 in Shopify, finds the fulfillment and tracking number, checks the carrier's status, and responds with a personalized update. If the order hasn't shipped yet, it explains the estimated processing time based on the product type.
Refund Playbook
Customer writes: "I want my money back for the broken mug." The AI identifies the customer's most recent order, checks if it contains a mug, verifies the order is within the return window, and either processes the refund (autopilot) or drafts the refund for approval (copilot). If the order is outside the window, it offers a store credit per the playbook's fallback instructions.
Cancellation Playbook
Customer writes: "Cancel my order please, I changed my mind." The AI checks fulfillment status. If unfulfilled, it cancels and confirms. If partially fulfilled, it explains which items can still be cancelled and which are already shipping. The playbook handles every branch.
Why This Matters
The shift from macros to playbooks is the shift from "canned answers" to "AI that actually solves problems." Playbooks give you the control of manual support with the speed and scale of automation. You write the strategy once, and the AI executes it across thousands of conversations -- each one slightly different, each one handled correctly.
Want to see playbooks in action? Try SupportPilot AI free -- default playbooks are configured automatically during setup.