Custom Skills

Your AI agent doesn’t just respond — it takes action. With Custom skills, you can set up workflows that trigger automatically as soon as the agent recognizes the right intent in a customer’s message. No manual effort. No missed opportunities.

Whether a customer is asking about a product, providing their contact information, or showing intent to buy, your agent can go beyond simple replies and carry out meaningful actions across your entire tech stack — delivering real, measurable results.

How does it work?

Custom skills can be triggered automatically when your AI agent detects a specific intent in a conversation or matches predefined keywords in a user’s message. Depending on how the skill is configured, triggers can also be based on the page the visitor is viewing, their location, or the time of day or time zone.

You define the steps and conditions, and the AI agent handles execution

What does the skill know about the customer?

Each Custom skill operates with a rich set of customer context, enabling your AI agent to make smarter decisions and execute more relevant actions. When a skill is triggered, it has access to detailed information about the customer, including:

• Name — Customer name.

• Email — Customer email address.

• Phone — Customer phone number, including country code. 

• Location — Customer location (e.g., San Jose, California, United States). 

• Timezone — Customer timezone (e.g., America/Los Angeles). 

• Last message — Most recent customer message. 

• Chat messages — Full conversation history in the current chat. 

• Current page — Page or URL the customer is viewing. 

• Browsing history — The last five pages the customer visited. 

• Tickets count — Total number of customer tickets, including status (open, archived, trashed, or spam). 

• Total orders — Total number of orders placed by the customer (available via Shopify integration). 

• Last chat date — Date and time of the most recent chat interaction. 

• Last ticket date — Date and time the last support ticket was created. 

• Last ticket comment — Most recent ticket comment with timestamp. 

With this level of insight, Custom skills can go far beyond basic automation. They allow your agent to personalize interactions, adapt workflows to each customer’s situation, and take actions that are timely, relevant, and aligned with the customer’s journey.

Actions

Every custom skill, besides being able to perform task-specific actions, can natively perform two built-in actions.

Send Message

This action allows the AI agent to communicate directly with the customer.
You provide the necessary context (for instance, discount code, order status, or product details), and the AI automatically generates a natural, on-brand response.

Collect Details

This action pauses the workflow and allows the AI agent to gather the required information from the customer.
You specify what data is needed (for example, order number, email, preferred date), and the AI agent handles the conversation until all details are collected, then continues the process.

Task-Specific Actions

In addition to built-in actions, the AI agent can perform task-specific operations related to chats, tickets, and customers. These actions enable the AI agent to, for example, update ticket statuses, tag or transfer chats, detect the language, and much more. Below is a list of all the actions your AI agent can perform.

Customers

Chats

Tickets

General

Templates and examples

Skills examples

You can get started faster with ready-made skill templates built around common support and sales workflows.

Each template includes a ready-to-use prompt that you can customize to fit your use case. To use a template, go to Custom skills, find the Skill examples section, and click the + icon next to a template. The content will appear in the editor, where you can modify it before saving your skill.

• Checkout Coupon Issuer — Offer a discount at checkout to reduce cart abandonment.

• Bundle Discount — Encourage larger purchases with multi-item deals.

• Product Insurance Upsell — Promote add-ons like warranties to increase revenue.

• Consultation Booking — Offer demos to interested customers, collect their details, and create a follow-up ticket for booking.

• Chat Supervision — Add a teammate as a supervisor to product-related chats to support sales.

• Abuse Detection — Detect abusive language and automatically close harmful chats.

• Instant Transfer — ​​Route chats to the right team as soon as predefined visitor variables are recognized, such as the customer’s language.

• Advanced Transfer — Route chats based on the customer’s intent.

• After-Hours Support — Capture requests outside business hours and create follow-up tickets.

Once you’ve edited the skill prompt, click Create skill to continue.

Review and save the skill

After you use a skill template or write your own prompt, review the generated workflow in the skill editor. You can adjust the skill name, edit the prompt, and check the steps the builder created. 

Use the preview on the right to see how the skill will work in a conversation. When you are ready, click Save (1) to keep the skill as a draft, or Save & activate (2) to save it and make it available to your AI agent right away.

Real-world examples

You can also create your own custom skills using available actions. Below are examples of how custom skills can support specific scenarios.

Route high-value inquiries to the right expert

What you’d like to achieve: Automatically detect high-value product inquiries, add a sales supervisor to the conversation, and tag the chat as sale so your team can respond faster, track opportunities, and improve conversion rates.

The prompt you could use: When a customer asks about the [product name], add a ‘Sales’ team supervisor to the chat and tag the conversation as sale.

Actions used:

• Supervise chat

• Tag chat

Collect missing details before creating a ticket

What you’d like to achieve: Automatically collect missing information, such as an order number, before creating a ticket so your team receives complete requests and can resolve issues faster.

The prompt you could use: When a customer reports an issue without providing their order number, ask for the order number in a natural way. Once they share it, create a ticket with the issue summary and order number, then tag the chat as support.

Actions used:

• Collect customer details

• Create ticket

• Tag chat

Review skill runs

Use the Runs section in the top-right corner of the custom skill editor to review how your custom skill performed in real conversations. It shows a log of each time the skill was triggered, including the date, time, and chat ID where it ran. 

Select a run to view the related chat, review the steps the skill followed, and check that each action was completed as expected. This helps you trace skill activity back to specific conversations and troubleshoot your configuration more easily.

How custom skills support your business

Custom skills let your AI agent take action across your tools automatically based on customer context. This includes not only what the customer says, but also details such as who they are, where they’re located, what page they’re viewing, and when the conversation started.

This helps your team respond faster, automate workflows, and deliver more relevant, consistent customer experiences. For example, you can trigger discounts during specific times of day, such as happy hour.

• Automate actions based on what your customers say.

• Connect your AI agent with the tools your business already uses.

• Route chats, update data, issue coupons, and automatically create tickets.

• Resolve customer requests faster and deliver more consistent service.

• Reduce repetitive manual work for your team.

• Build workflows in plain language without manual setup.

Ready to get started?

Just head to the Team → AI agents → Skills to build custom skills for your sales and support workflows.