Introducing Sibbi: Conversational Commerce Built on Commerce Superintelligence

Introducing Sibbi: Conversational Commerce Built on Commerce Superintelligence

Shopping online has always required the shopper to do most of the work. Type a keyword. Scroll through a grid. Filter by size, color, price. Click into a product page. Go back. Try a different query. Repeat until something looks right or patience runs out.

Sibbi changes that. It is the conversational interface of Marqo's Commerce Superintelligence, an autonomous agent that guides shoppers from discovery through post-purchase using deep product understanding.

A shopper types "I need running shoes for flat feet under $150." Sibbi understands the intent, asks whether they prefer road or trail, and surfaces relevant products from real inventory. Another shopper uploads a screenshot from Instagram. Sibbi finds matching products in the retailer's catalog and refines with "same style but in black." When a shopper adds a jacket to cart, Sibbi recommends the pants and boots that complete the outfit. The shopper selects their size, adds to cart, and checks out without leaving the conversation.

This is not a chatbot with a product feed attached. Sibbi is a commerce agent trained on each retailer's specific catalog, grounded in real inventory, and capable of handling the full shopper journey from first question to post-purchase.

Why Conversational Commerce Matters Now

The way shoppers interact with online stores is shifting. Consumers who use AI assistants daily now expect the same intelligence when they shop. They expect to describe what they want in plain language, show an image of something they like, ask follow-up questions, and get useful answers, not a grid of loosely related products.

Most tools built to meet this expectation are generic. They are general-purpose language models pointed at a product feed. They sound fluent but they do not actually know the catalog. They recommend products that are out of stock, describe features that do not exist, or fail to understand the difference between categories specific to one retailer's assortment.

Conversational commerce only works when the agent doing the conversation actually knows the merchandise. That requires something most platforms cannot deliver: deep product understanding, specific to each retailer, grounded in real inventory, and continuously refined by real shopper behavior.

What Sibbi Can Do

Guided Discovery

Sibbi interprets what shoppers mean, not just what they type. When a shopper asks for "something elegant for a garden party," Sibbi understands the occasion, the aesthetic, and the implicit constraints. It can ask clarifying questions: "Are you looking for a dress, a jumpsuit, or separates?" "Do you have a color preference?" Each response narrows the results toward what the shopper actually wants.

This matters because the highest-value discovery queries are the ones where shoppers search by intent rather than exact product terms. Style-based queries, use-case queries, and incomplete descriptions are where the revenue opportunity is largest, and where keyword search and behavioral ranking fall short. Sibbi handles these queries natively because it understands products, not just keywords.

Visual Search

Shoppers increasingly discover products through images: a screenshot from social media, a photo of something a friend wore, a picture of a room they want to recreate. Sibbi accepts image inputs and finds matching or visually similar products from the retailer's catalog.

What makes this different from standalone image search tools is that Sibbi combines visual and text signals in the same conversation. A shopper can upload a photo and add "but in a warmer color" or "similar silhouette but shorter length." The visual and semantic signals are processed together, not as separate queries stitched together after the fact.

Cross-Sell and Complementary Products

When a shopper shows interest in a product, Sibbi understands what goes with it. Not based on what other shoppers happened to click in the same session, but based on genuine product relationships: what complements the style, what completes the outfit, what enhances the purchase.

A shopper adding a navy blazer to cart sees recommendations for trousers, shirts, and shoes that actually work together visually and stylistically. A shopper buying a moisturizer sees the serum that pairs with it. These recommendations are grounded in product understanding, not collaborative filtering, which means they work for new products and long-tail items that have no co-purchase history.

Add to Cart

Sibbi does not just surface products. It closes the loop. A shopper can select size, color, and quantity and add to cart within the conversation. The path from discovery to cart stays inside a single experience with no redirects, no friction, and no context loss.

Post-Purchase

Most commerce AI stops at checkout. The intelligence that helped the shopper find the right product disappears the moment they buy it. Order tracking goes to a separate system. Returns go to a support ticket queue. The relationship between the shopper and the brand's intelligence layer is severed.

Sibbi extends beyond the purchase. The same agent that helped a shopper discover and buy a product can answer "where is my order?", process a return, or suggest a complementary product for their next purchase. Post-purchase interactions become new opportunities for discovery, not dead-end support tickets.

One agent, one conversation, from first query to post-purchase.

How Sibbi Is Different: Built on Commerce Superintelligence

Sibbi is the first agent built on Marqo's Commerce Superintelligence, a single intelligence layer that combines deep product understanding with behavioral data and personalization to power every shopper interaction.

What this means in practice:

Trained on each retailer's catalog. Sibbi is not a generic model that works the same for every store. Marqo trains a dedicated AI for each retailer that understands their specific products, their customer vocabulary, and their commercial priorities. The model that powers Sibbi for a luxury jewelry brand is fundamentally different from the model that powers Sibbi for a sneaker marketplace.

Grounded in real inventory. Every product Sibbi recommends is verified against the retailer's live inventory. Products that are out of stock are automatically excluded. Prices, attributes, and availability are current. This eliminates the hallucination problem that plagues generic AI agents. No phantom products. No fictional attributes. No recommendations for items that cannot be purchased.

Commercial intelligence built in. Sibbi does not just surface relevant products. It surfaces the right products for the business. Margin, inventory priority, seasonal strategy, and promotional objectives are embedded in how Sibbi ranks and recommends. When two products are equally relevant to the shopper, Sibbi can prefer the one that drives more value for the retailer, without requiring manual merchandising rules.

Same intelligence across every touchpoint. Sibbi runs on the same AI that powers Marqo's search, merchandising, and recommendations. Improving the model once improves every surface. The product understanding that helps Sibbi answer a conversational query is the same understanding that ranks search results and generates recommendation carousels. One intelligence layer, consistent across the entire commerce experience.

Results From Commerce Superintelligence

The intelligence layer that powers Sibbi is already delivering measurable results in production at some of the world's largest retailers.

Fashion Nova attributed $130 million in incremental revenue to its Marqo implementation. Mejuri saw a 19.8% increase in search-driven conversion. KICKS CREW achieved a 17.7% conversion rate improvement. Kogan generated $10.1 million in attributable revenue impact. SwimOutlet went from initial integration to live production A/B testing within five days and saw a 10.6% increase in search add-to-cart rate.

Sibbi brings this same proven intelligence into a conversational experience. Commerce Superintelligence enables measurable improvements across every dimension of conversational commerce: higher close rates because the agent recommends products grounded in real inventory and real relevance, better clarification quality because the agent understands products deeply enough to ask the right follow-up questions, and smarter product presentation because the system knows when a shopper's intent is clear enough to surface results versus when to ask for more context.

Getting Started

Sibbi deploys with a single line of code. There is no months-long integration project. Retailers can have Sibbi live and interacting with shoppers quickly, and the platform is designed to surface measurable results within 14 days.

The onboarding process starts with catalog ingestion. Marqo's dedicated AI is trained on the retailer's specific catalog. Once the model is ready and the inventory connection is live, Sibbi is available across any surface where the retailer wants to offer conversational commerce: product pages, category pages, search, or a dedicated shopping assistant experience.

For retailers already using Marqo's Commerce Superintelligence for search and recommendations, Sibbi is a natural extension. The same intelligence that already powers discovery now powers conversation. For retailers new to Marqo, Sibbi is a complete introduction to what Commerce Superintelligence can do.

Frequently Asked Questions

Is Sibbi a chatbot?

No. Sibbi is a conversational commerce agent built on Commerce Superintelligence. The distinction matters. A chatbot is a general-purpose conversation interface. Sibbi is purpose-built for retail, trained on each retailer's specific catalog, grounded in real inventory, and capable of completing transactions and handling post-purchase interactions. It does not just answer questions. It drives commercial outcomes.

How does Sibbi avoid recommending out-of-stock products?

Sibbi is connected to the retailer's live inventory. Every recommendation is verified against current availability before being returned to the shopper. When a product goes out of stock, Sibbi automatically removes it from consideration. There is no manual curation step required.

Does Sibbi work for any retail category?

Yes. Because Sibbi is trained on each retailer's own catalog, it adapts to any category. The model that powers Sibbi for a fine jewelry brand understands gemstones, metals, and occasion-based gifting. The model for a sneaker retailer understands colorways, silhouettes, and limited releases. Commerce Superintelligence is the framework. Sibbi is the expression of that framework for any category of retail.

How does Sibbi handle post-purchase?

The same agent that helped a shopper discover and purchase a product can answer order status questions, process returns, and suggest complementary products for future purchases. Sibbi draws from real order data and the same product intelligence that powers discovery. There is no handoff to a separate support system.

How long does it take to see results?

Retailers can expect measurable results within 14 days of deployment. The platform combines immediate product understanding with ongoing behavioral refinement, delivering value from day one that compounds over time.

Sibbi and Commerce Superintelligence are available today for enterprise retailers. Book a demo to see what Sibbi can do with your catalog.