VWAM is an AI-native decision engine that guides consumers at the moment intent forms — directly on your owned digital properties.
Unlike traditional chatbots or search tools, VWAM does not rely on keywords, menus, or static flows. Instead, it holds a real-time conversation that understands what a customer is trying to accomplish and actively guides them toward the best next step, whether that’s discovering a product, learning about an offering, or exploring content.
VWAM is built to operate entirely within your brand’s control. Every response is grounded in your data, your rules, and your brand voice — ensuring accuracy, safety, and consistency across experiences.
Before you begin
Before implementing VWAM, determine:
Your primary business goal (conversion, engagement, education, loyalty, monetization).
Where VWAM will be embedded.
What rules and guardrails are required to protect brand voice and compliance.
Clear goals and defined data inputs improve recommendation quality and performance.
How brands use VWAM
VWAM can support multiple stages of the customer journey, depending on how it’s configured and where it’s embedded.
At a high level, brands use VWAM to:
Reduce friction in product discovery and research.
Increase engagement and time spent with brand content.
Drive conversions and qualified referrals.
Educate customers at scale using brand-approved content.
Capture rich first-party intent and preference data.
How VWAM is different from traditional chatbots
VWAM is not a generic support bot or a thin wrapper around an LLM.
Key differences include:
Brand-controlled intelligence
VWAM is grounded in your product feeds, content feeds, and business rules. The AI can only recommend or reference what you explicitly provide.
Guardrails and rules by design
Brands define constraints, policies, and logic that the AI must follow, preventing hallucinations, off-brand responses, or risky edge cases.
Embedded anywhere in the experience
VWAM is not limited to a floating support widget. It can be embedded inline on product pages, landing pages, editorial content, loyalty portals, or campaign experiences.
Purpose-built agents, not one generic bot
Brands can create multiple agents, each designed for a specific job (e.g., product finder, gift guide, stylist, trip planner).
VWAM use cases
Vertical | Use Case Description | Primary Goal |
E-commerce/Retail | AI Product Finder for Direct Sales: | Drive Purchase / Lift Conversions |
Entertainment/Media | Character Interaction/Brand Forward Experience: Creating an AI agent based on a fictional persona or influencer (e.g., "Pep Talk with Ted Lasso") to foster positive brand association. | Entertain / Engagement / Consideration |
Travel/Hospitality | AI Trip Planner/Itinerary Creator: | Engagement / Trip Ideation / Data Gathering |
Media/Editorial | Chat with the Editor/Expert: | Educate / Editorial Engagement / Drive Affiliate Sales |
Retail/E-commerce | Seasonal/Holiday Gift Finder: | Drive Seasonal Sales / Conversion |
Luxury/Beauty | Professional Stylist/Expert Chat: Positioning the AI as a professional stylist to offer beauty or hair recommendations, ensuring the chat follows specific brand guidelines and educational goals (e.g., product finder for hair care). | Educate / Product Consideration / Engagement |
Media/Sponsored Content | Sponsored AI Chat/Ad Solution: | Monetize Audience / Drive Affiliate Sales |
Retail/Loyalty Programs | Concierge Personal Shopper for VIPs: | Curation / Loyalty Program Engagement |
CPG/Awareness | Top-Funnel Awareness Play: | Brand Awareness / Awareness |
Data, ownership, and control
Brands retain ownership of all conversation data and insights.
Preference and intent signals collected during chats can be integrated into customer profiles via API.
Data is tenant-isolated and not shared across brands.
VWAM supports GDPR compliance and enterprise-grade privacy requirements.
VWAM gives brands a way to actively participate in customer decision-making — without giving up control to third-party platforms or generic AI tools.
By combining conversational intelligence, brand guardrails, and flexible deployment, VWAM enables brands to meet customers where intent forms and guide them with confidence.
Next steps
Create and configure your first agent.
FAQ
How do we ensure brand safety and control the Vwam content & voice?
Brands manage control through Rules and Guardrails. The AI is "grounded" by having the brand upload a product feed or content feed (CSV, database, or eventually uploaded documents like blog posts) that the LLM must pull from when making recommendations, ensuring content is vetted and accurate. Custom rules can be written to prevent edge cases, keep within specific parameters.
What if our product data isn't perfectly structured or indexed? Will the AI still work?
Yes, the power of AI is that it can understand data even if it is not perfectly structured. The tool can be trained on files like CMS data or blog posts (a feature currently being implemented) rather than just structured product feeds, which is useful for educational or engagement-focused applications.
Can we use VWAM to collect data and integrate it with our customer profiles?
Yes, the brand retains ownership of all chat transcripts and conversation data. Data points collected during the chat (e.g., preferences, interests, price range) can be extracted and appended directly to the customer profile in the central database (CDP) via API integration.
Can we create multiple different Vwam Experiences?
Yes, brands are moving away from building one general AI experience. They can create multiple purpose-specific agents for different contexts, such as a holiday gift finder, a roadside assistance plan expert, or a general product assistant, with each agent having its own specific role and rules.
How does the recommendation logic currently work, and how will it integrate with our CDP?
In the short term (beta), the chat leverages the Large Language Model (LLM) autonomously to make basic decisions based on the conversation and the content feed provided. Full integration with the Blue Conic CDP decisioning engine (for "next best action modeling") is estimated to be 4 to 6 months away. Custom rules to inform product recommendations can be created within the “Rules” engine
How are attribution and conversion tracked for partner referrals or site clicks?
For partner referrals to external sites, the recommended path is to build the partner feed using URLs that include a UTM tag. This allows tracking of referral traffic and gives credit to the brand.
What if we have a massive, dynamic product feed (thousands of retailers/changing inventory)?
Currently, feeds are uploaded manually (e.g., via CSV). However, implementing API connections to commerce platforms to handle massive, dynamic product feeds (where inventory changes by the hour) is on the roadmap for the medium term.
Does the LLM have set rules / restrictions to block bad stuff / inappropriate content (such as a list of terms that the chatbot does not entertain and instead stops).
There is currently no blacklist functionality. But a simple or detailed rule could solve a brand's challenges. They could add their entire security policy into a rule if they want to.
What LLM do you use?
VWAM uses a variety of models from OpenAI and Anthropic. In the future we will enable customers to leverage their own model.
Is data shared across the platform / other brands?
No, all data resides within the tenant.
Is this compliant in EMEA?
Yes we are GDPR compliant