Playbooks allow you to provide specific guidance to your alby AI Agent on how to handle different customer scenarios, such as product returns, helping them shop for the right item or recommending top selling items.
Playbooks can ensure your alby Agent delivers consistent, personalized experiences that align with your brand requirements. Playbooks are sets of instructions for alby on how to deal with specific scenarios for customers. One or more Playbooks can be activated for each Agent.
Think of Playbooks as enhanced rules for alby. Like rules, they have two parts: a trigger situation that activates the Playbook, and instructions that tell alby how to respond.
How Playbooks Work
Each Playbook consists of two main components:
Situation: The specific scenario or context when the Playbook should be activated. Examples:
A customer is asking about order status
A shopper is browsing specific product categories.
Instructions: Detailed guidance for how the Agent should handle the situation, including which Actions to take and in what order
The below examples show how Playbooks let you create tailored, context-specific shopping experiences for customers while keeping your rules and branding consistent. Some of the use cases for Playbooks in your brand include:
Example one: Customize how alby handles product recommendations for specific categories.
Playbook name:
Recommend skis and snowboards
Situation:
Customer asks about products in a specific category, such as winter sports equipment
Product Search
Compatibility Check
Best Sellers
Product Finder
Instructions:
Step 1: Clarify their preferences, skill level, and intended use if not already provided.
Step 2: Use /ProductFinder to help the customer narrow down suitable ski or snowboard options based on their needs.
Step 3: Use /ProductSearch to present a selection of relevant skis or snowboards.
Step 4: Once the customer expresses interest in a specific ski or snowboard, use /BestSellers to find the best bindings for the selected product.
Step 5: Address any further questions, adjust recommendations based on customer feedback, and confirm the final selection.
Step 6: Summarize the selected products and ensure the customer is satisfied with their choices before concluding the interaction.
Example two: Offer support with projects to create comprehensive product recommendations
Playbook name:
Project-based shopping assistance
Situation:
Customer browses a home improvement category
Actions Available:
Product Search
Product Finder
Instructions:
Step 1: Gather detailed information from the user about their project, including goals, preferences, requirements, and any specific products or styles they are interested in.
Step 2: If the user is unsure about what they need, use /ProductFinder to guide them through a series of questions to clarify their needs and narrow down suitable product categories.
Step 3: Use /ProductSearch to search the product catalog based on the user's specified criteria or the results from /ProductFinder.
Step 4: If the user wants to compare multiple products, use /Comparison to evaluate options based on the criteria most important to the user (e.g., price, features, style, compatibility). Present a clear summary of the comparison.
Step 5: Present a comprehensive set of recommendations to the user, including main products and any suggested matching or complementary items. Clearly explain the rationale for each recommendation.
Step 6: Ask the user for feedback or if they need further refinement. If so, repeat the relevant steps (e.g., adjust search criteria, compare additional products, or find new matching items) until the user is satisfied.
Step 7: If the user has additional project requirements or new goals, return to Step 1 and repeat the process as needed.
Tips for Playbooks
Keep instructions clear and specific with exactly what you want your Agent to do.
Duplicate and test your Playbooks in the playground before adding them to your widgets or chat.
Use the Inbox to review customer interactions, and update your Playbooks accordingly.
Keep each Playbook’s use focused — only use one Playbook per topic.
Be creative — Playbooks provide a sandbox for exploring your alby Agent's full capabilities.
Prerequisites
Before you make a Playbook, you should decide what Agent will use the Playbook, and what Actions it will use.
Create a Playbook
In the alby dashboard, click
Agents.
Navigate to the Playbooks section.
Click Create New Playbook.
Name your Playbook.
- The name is included when alby is determining when to use this Playbook, so name it something descriptive about the situation.
Define the situation that triggers this Playbook.
Write your instructions for the Agent, incorporating available Actions using the /.
Click Save.
Testing your Playbooks
In the playground of the Agents page, you can test your Agent and any Playbooks to make sure they work as expected.
Type your prompt into the chat window on the Playbooks page and see if your Agent runs the Playbook as you’d expect it to. If it doesn’t, edit either the trigger or instruction information in the Playbook, then retry.
For example, you could test the first example from above by asking about a ski or snowboard and seeing if alby follows the Playbook.
The preview will only test your live Playbooks. To avoid affecting active widgets or chat, create a test Agent that isn't connected to any widgets or chat, and set your Playbooks live for testing. Otherwise, you might update settings on live widgets or chats.