Creating a chat agent in VWAM is fast and intuitive. In just a few steps, you can generate a fully configured AI-powered experience — with optimized recommendations, brand alignment, and structured logic built in from the start.
An agent is your conversational experience inside VWAM. It interacts with consumers, recommends products, and responds based on rules you define.
You can configure each agent across four areas:
Design – Visual styling and layout
Content – Text, media, and interface copy
Collections – Product feed and product display logic
Rules – Recommendation logic, conversation behavior, and toneBefore you begin
Before you begin
Before creating an agent, ensure:
If you plan to recommend products from a feed, ensure your product feed is uploaded and available
You understand your deployment type:
On-page – Embedded in your site.
Owned web – Standalone page.
You can test your agent in the Editor at any time during setup. Publishing is not required to begin testing.
Create and configure an agent
From the Dashboard, click + Create Agent.
In the creation modal, enter:
Objective – Describe your brand and goal. This is the most important input — clear guidance around audience, goals, tone, and constraints allows VWAM to generate a highly relevant and optimized workflow.
For example: “Help my customers find the best running shoes at Adidas (www.adidas.com). Let’s ensure we really understand the experience of the runner and how frequently they run, while keeping budget in mind.”Deployment Type – Select On-page (embedded in your site) or Owned web (standalone URL).
Click Create.
Clear guidance around audience, goals, tone, and constraints will allow the AI to generate a highly relevant and optimized workflow.
What happens next
Once created, VWAM automatically:
Ingests your brand style guide from your website (if a URL is provided)
Builds rules aligned to your objective
Maps recommendations against your selected product catalog
Aligns responses with your brand voice
In approximately 15 seconds, your agent is ready to edit — with a structured workflow, brand voice, guardrails, and product-aware recommendation logic already in place.
Prefer to build manually? Leave the Objective field blank to configure your agent from scratch without AI automation.
Once your agent is created, you are redirected to the Editor.
Configure design settings
Open the Design tab to control visual styling and layout.
You can configure:
Color – Backgrounds, text, and interface elements.
Typography – Font family and appearance.
Alignment – Positioning of interface elements.
Style –
Bubble – Chat appears like text messaging.
Plain – Chat appears without messaging bubbles.
Size – Adjust sizing of various interface elements.
Use these settings to align the agent with your brand guidelines.
Add brand content and media
Open the Content tab to personalize text and upload media. This allows you to further personalize with uploaded imagery and customized agent phrasing
Text settings
Configure the following fields:
Title – Heading text.
Loading – Text shown while the AI is generating a response.
Input – Default placeholder text in the chat input field (goes away when a user starts typing into it).
Disclaimer – Text displayed at the bottom of the agent.
Media settings
Upload brand assets:
Logo – Displays in the header.
Avatar – Appears next to the agent’s messages.
Send – Icon used in the chat input.
Background – Chat background image.
Connect a product feed
A product feed is required only if you want the agent to recommend products from your catalog. Agents can run without one.
Open the Collections tab.
1. Select a feed
Use the Feed dropdown to select the appropriate product feed.
All uploaded feeds appear in the dropdown.
2. Configure product display
Under Display, control how product cards appear:
Drag the dots to reorder product information.
Test display by asking the agent to show a random product.
Adjust order until it matches your preferred layout.
3. Formatting options
Max Products – Limit the number of products recommended.
Line Limit – Limit the number of lines shown in product cards.
Welcome Products – Display products before a user starts chatting.
Configure rules
Rules define how the agent responds and what it recommends.
Open the Rules tab to control AI behavior.
Recommendations
Control which products appear and under what logic.
Example: Always include a specific best-selling product alongside other recommendations.
Conversation
Define how the agent responds in specific scenarios.
Example: Direct consumers to a support page when they ask about refunds.
Tone of voice
Provide guidance on how the agent communicates.
Examples:
Professional and courteous
Playful and energetic
For advanced configuration, refer to the dedicated Rules guide.
Test your agent
Test your agent directly in the Editor.
The chat is live.
Interact with the agent regularly while building.
After adding or adjusting rules, validate that behavior reflects your updates.
Refine rule phrasing as needed to achieve the desired outcome.
You do not need to complete configuration before testing. Build and test iteratively.
Next steps
Deploy and publish your agent.
FAQ
Can I create multiple agents?
Yes. You can create multiple agents for different use cases.
Can I duplicate an agent?
No. Duplication is not supported at this time.
Can I test without publishing?
Yes. You can fully test within the Editor before publishing.
Can I change the product feed later?
Yes. You can switch feeds within the Collections tab.
Do I need to configure everything before testing?
No. You can iteratively configure and test as you build.