Skip to content

Glossary

Webscale AI uses one set of terms for its products, services and the market they serve. The core terms are the Commerce Context Engine, the AI Shopping Assistant, Audience Segmentation and Forward Deployed Engineers, each defined on this page.

This glossary defines the terms Webscale AI uses for its products, its services and the market they serve. Our team, our partners and these docs use each term the way this page defines it.

Labels in bold match the Webscale AI console exactly. Each term has its own link, so you can point anyone straight to a definition.

Name What it is
Commerce Context Engine The data layer under every Webscale AI product
AI Shopping Assistant The assistant that sells and serves on the merchant’s storefront
Audience Segmentation Live audiences built from plain-English descriptions
Full Platform Both products on a merchant-owned orchestration layer
Forward Deployed Engineers The commerce data experts who get a merchant’s data ready
Pilot A trial run of the AI Shopping Assistant on the merchant’s store, under a pilot agreement

Webscale AI is the commerce AI brand of Webscale Networks. It covers the Commerce Context Engine, the AI Shopping Assistant, Audience Segmentation and the Forward Deployed Engineers who deliver them. Its home is ai.webscale.com.

Webscale Networks is the parent company of Webscale AI. It has run commerce infrastructure since 2013.

The Agentic Commerce OS is how Webscale AI describes its whole stack. Merchants capture first-party data, refine shopper intent and act on the storefront, all on one layer the merchant controls.

Digital GTM team is how Webscale AI describes Lagrange. GTM stands for go-to-market. Lagrange’s AI capabilities take on routine work a sales, service and marketing team does: helping shoppers choose, answering routine questions and reaching new shoppers. They are built to handle more conversations without adding staff at the same rate. Plans differ in what they cover. Order and returns management comes with Pro + Ads.

A merchant is a business that sells online and uses Webscale AI.

A shopper is a merchant’s customer: the person who talks to the AI Shopping Assistant or lands in a segment.

The Commerce Context Engine is the knowledge foundation behind Lagrange. It gathers a merchant’s product, pricing, policy and other business information, keeps it current and gives the AI the right facts to answer from. Every plan runs on it. It is never sold on its own.

Model-ready data is commerce data that is structured, validated and complete enough for an AI agent to read and trust without guessing. The Commerce Context Engine produces it.

First-party data is data a merchant collects directly from its own shoppers: browsing, carts, orders and profiles.

A grounded answer comes from the merchant’s own live data and sources, not from a language model’s general knowledge. The AI Shopping Assistant is built to give grounded answers.

A connector is a built-in link between Webscale AI and another system, such as a commerce platform, a helpdesk or a marketing tool.

Catalog sync is the scheduled job that brings a merchant’s products into the Commerce Context Engine. The console shows its status and its schedule.

Product coverage is the console report on how complete a merchant’s synced catalog is. It runs data-quality checks and rates each issue by severity.

Data readiness is how complete a merchant’s catalog, attributes and documentation are for AI agents to answer from. It is also the name of the third stage of a pilot.

The merchant-owned orchestration layer means the merchant controls the rules, guardrails and knowledge its AI agents follow. It is part of the Full Platform.

Plans and prices are on the pricing page.

The AI Shopping Assistant is one conversational interface on the merchant’s storefront that sells and serves. It handles product discovery, product questions, orders and returns in a single thread, grounded in the live catalog.

An agent is one configured AI Shopping Assistant in the console, with its own name, design, instructions and install tag. One account can run several agents.

Specialist agents are purpose-built agents, such as product discovery, that the AI Shopping Assistant routes each part of a question to. The console lists them as sub-agents under Advanced settings.

Audience Segmentation builds live audiences from plain-English descriptions on a merchant’s first-party data. It keeps each audience current as shoppers act and syncs it to the merchant’s marketing tools. Its earlier name was AI Segmentation.

A segment is one audience built in Audience Segmentation.

The Full Platform brings the AI Shopping Assistant and Audience Segmentation together on a merchant-owned orchestration layer. It is priced for each merchant’s stack.

A pilot is a trial run of the AI Shopping Assistant on the merchant’s store, under a pilot agreement with Webscale. Webscale AI confirms fit before setup starts.

A shopper conversation is the unit the AI Shopping Assistant is billed by.

The AI readiness scanner is a free tool at ai.webscale.com/ai-readiness. It scores a store from 0 to 100 on how readable it is to AI shopping agents and answer engines, then lists the top fixes.

These are the measures Webscale AI uses to describe what the AI Shopping Assistant does for sales and for service.

An assisted conversion is a purchase that follows a shopper’s conversation with the AI Shopping Assistant. It shows how often the assistant helps a shopper reach a buying decision. It is the lead sales metric for the assistant.

Cost per resolution is what it costs on average to resolve one customer issue, such as a product question or a return request. It is the total cost of handling issues divided by the number of issues resolved. Resolving routine issues in the conversation lowers it. It is the lead service metric for the assistant.

Self-service resolution rate is the share of customer issues resolved in the conversation, without a person from the merchant’s team stepping in. A higher rate means fewer tickets for the team.

Forward Deployed Engineers are commerce data experts who work inside a merchant’s team. They find the data AI agents cannot answer from today and map it into the Commerce Context Engine.

Forward Deployed Engineering is the service Forward Deployed Engineers deliver, from the first data review through launch.

A pilot moves through five stages, in this order.

Stage What happens
Kickoff One session on the merchant’s catalog, business rules and what shoppers ask
Expert interviews Time with the people who answer shoppers’ questions today, capturing what they know
Data readiness work The catalog is structured, attributes are fixed, documentation is ingested and configuration is tuned
Testing and acceptance Test cycles against real shopper questions, with the merchant’s sign-off before launch
Launch Live on the storefront, on the date set at kickoff

The console is the merchant’s admin for Webscale AI, at dashboard.ai.webscale.com.

Prompt instructions are the rules a merchant writes for an agent, such as brand voice, business rules and what to say or avoid.

Trainings are folders of sources an agent learns from. A folder can hold files, text snippets, webpages and product documents. A product document applies only to the products it is linked to.

The Knowledge base holds a merchant’s own term definitions for Audience Segmentation to use when it builds segments. It is separate from Trainings, which feed the AI Shopping Assistant.

Custom workflows are flows a merchant builds from blocks on a canvas. A workflow can collect details, branch, call a webhook, ask an AI model, show a message, create a helpdesk ticket, suspend a visitor or hand the conversation to a person.

A trigger is the block that starts a custom workflow. It fires on keywords or as a hidden trigger with no visible keyword.

The Playground is the console panel where a merchant chats with an agent to test a change before shoppers see it.

Install tags are the script snippet that puts an agent on the merchant’s storefront.

The widget is the chat window shoppers see on the storefront. Merchants style it on the Design tab.

Logs are the record of every conversation an agent has, with a preview of each one and shopper feedback.

A live agent handoff passes a conversation from the AI Shopping Assistant to a person, through a custom workflow or a connected helpdesk.

The segment library is a set of pre-built segment templates for revenue, lifecycle, engagement and behavior.

Anonymous visitor tracking is the data setting that controls whether Webscale AI tracks visitors who have not identified themselves.

Lagrange runs on Shopify, Shopware and Magento. Discovery and Growth run on Shopify only. On Shopify, Lagrange is a marketplace app, installed from the Shopify App Store. Pro + Ads also runs on Shopware and Magento (Adobe Commerce and Magento Open Source), through a supported integration set up with our team. Their marketplace apps are planned.

Adobe Commerce is the commercial edition of Magento, sold by Adobe.

Magento Open Source is the free edition of Magento. These docs call it Magento after the first mention.

A helpdesk connector links the AI Shopping Assistant to the helpdesk a merchant already uses, such as Zendesk. It opens tickets and receives live agent handoffs.

A marketing destination is a tool Audience Segmentation sends segments to, such as Klaviyo.

Webeyez is a Webscale AI partner. Its friction detection triggers the AI Shopping Assistant in the same session, when a shopper is about to give up.

Agentic commerce is shopping where an AI agent finds, compares and buys products for a person. Read more on agentic commerce.

An AI shopping agent, such as ChatGPT or Gemini, acts for a shopper. It reads stores to recommend or buy products. It is different from the AI Shopping Assistant, which runs on the merchant’s own storefront.

An answer engine is an AI system that answers a question directly instead of listing links. ChatGPT, Perplexity, Gemini and Google AI Overviews are answer engines.

AEO, or answer engine optimization, is the work of making pages easy for answer engines to find, quote and cite.

MCP, the Model Context Protocol, is how an AI agent calls a tool that someone publishes instead of reading their web page.

UCP, the Universal Commerce Protocol, is an open specification for agent-led commerce transactions, co-developed by Google and Shopify.

ACP, the Agentic Commerce Protocol, is an emerging standard for how catalogs, carts and checkout are exposed to AI agents.

AP2, the Agent Payments Protocol, is a protocol for payments an AI agent makes on a shopper’s behalf.

A CDP, or customer data platform, is a system that unifies customer profiles from many sources, usually for marketing. Many merchants already run a CDP.

A PIM, or product information management system, is where some merchants manage their product data before it reaches the storefront.

An ESP, or email service provider, is the platform a merchant sends marketing email from, such as Klaviyo.

Frequently asked questions

What is the difference between the AI Shopping Assistant and an AI shopping agent?
The AI Shopping Assistant is the Webscale AI product. It runs on a merchant's own storefront and answers from that merchant's live data. An AI shopping agent, such as ChatGPT or Gemini, acts for a shopper. It reads many stores to recommend or buy products.
How is the AI Shopping Assistant different from a traditional chatbot?
A traditional chatbot follows scripts and sends hard questions to a help center. The AI Shopping Assistant answers from the merchant's live catalog and data. It handles product discovery, orders and returns in one conversation.
What does a Forward Deployed Engineer do?
Forward Deployed Engineers are commerce data experts who work inside a merchant's team. They find the data AI agents cannot answer from today and map it into the Commerce Context Engine.