BuyWhere
← Back to documentation

agent-commerce-glossary

Agent Commerce Glossary

By Draft · April 2026

Key terms for understanding AI-powered shopping agents and product catalog infrastructure.

15 Key Terms

MCP (Model Context Protocol)

A standardized protocol for AI models to interact with external tools. MCP lets AI agents discover and use tools — like a shopping catalog — without custom integration code for each tool.

Example: An AI agent using BuyWhere via MCP doesn't call an HTTP endpoint; it calls the search tool directly, and the MCP server handles the translation.

GEO (Generative Engine Optimization)

The practice of optimizing content so AI agents and generative search engines can discover, understand, and cite it. Similar to SEO but targeting AI crawlers rather than search engine spiders.

Example: Writing product descriptions that AI agents can parse and cite, or maintaining an llms.txt file for AI crawler discovery.

Agent-Native

Design philosophy where a product, API, or service is built specifically for how AI agents operate — including tool-calling interfaces, structured data outputs, and batch operations optimized for agent workflows.

Example: BuyWhere's batch-lookup endpoint, which lets an agent compare multiple products in a single call rather than making dozens of individual requests.

Structured Product Data

Normalized, typed product information (price, availability, SKU, specifications) in a consistent format across all retailers. Enables cross-retailer comparison.

Example: Two retailers might list "Sony WH-1000XM5" differently, but BuyWhere normalizes both into {brand: "Sony", model: "WH-1000XM5", ...} so an agent can compare them directly.

Price Comparison API

An API endpoint that returns a single product's price across multiple retailers, sorted by price. Often includes price rank, savings vs. average, and buybox price.

Example: GET /agents/price-comparison?product_name=Sony+WH-1000XM5 returns prices from Shopee, Lazada, Amazon SG, etc., ranked from cheapest to most expensive.

Catalog API

An API that provides access to a product database — search, filter, browse by category, look up by ID. The foundational data layer for shopping agents.

Example: GET /agents/search?q=laptop&category=Electronics returns a list of laptops matching the query with prices and availability.

Merchant Coverage

The number and variety of retailers (merchants) whose products appear in a catalog. Higher coverage means an agent can find more options and give better recommendations.

Example: BuyWhere covers 50+ merchants across 7 markets (Singapore, Philippines, Indonesia, Vietnam, Malaysia, Thailand, US), from major platforms like Shopee to specialty retailers like Challenger.

Product SKU (Stock Keeping Unit)

A unique identifier for a product within a retailer's catalog. SKUs vary by retailer — the same product has different SKUs on Shopee vs. Lazada. Cross-retailer product matching normalizes these.

Example: A Nike shoe might be SKU "AJ2024-001" on Nike's site, "NKE-AJ2024" on Amazon, and "12345678" on Shopee. BuyWhere's canonical SKU links them all.

Price Freshness

How recently a product's price was last updated. Critical for agents making purchasing recommendations — stale price data could direct users to unavailable or overpriced items.

Example: BuyWhere tracks data_updated_at on each product. Agents can filter or flag results where freshness exceeds a threshold (e.g., 24 hours).

LLM Tool Call

A direct invocation of a function or tool by a Large Language Model. When an agent decides to search for products, it makes a tool call rather than generating text.

Example: An agent reasoning: "I need to find headphones under $100" → tool call to search(max_price=100) → receives product list.

JSON-LD (JSON for Linked Data)

A structured data format using JSON that search engines and AI systems can parse. Products with JSON-LD markup are more likely to be correctly parsed by agents.

Example: A product page with <script type="application/ld+json"> containing @type: Product, name, price, availability that BuyWhere generates automatically.

Schema.org

A vocabulary standard for structured data on the web. Defines types like Product, Offer, PriceSpecification that enable rich product markup for SEO and AI discovery.

Example: BuyWhere's product schema uses Schema.org vocabulary: { "@type": "Product", "name": "...", "offers": { "@type": "Offer", "price": "99.00", "priceCurrency": "SGD" } }.

llms.txt

A text file at a website's root (/llms.txt) that provides AI agents and crawlers with a structured overview of available content and API endpoints. Similar to robots.txt but for AI systems.

Example: BuyWhere's llms.txt lists: catalog coverage (1M+ products, 50+ merchants), API endpoints, authentication requirements, and links to documentation.

Semantic Search

Search that understands meaning and context, not just keyword matching. In product search, this means queries like "what's a good laptop for coding" return relevant results rather than just products with "laptop" in the title.

Example: A semantic search for "quiet washing machine" returns models with low noise ratings, not just products with "quiet" in the description.

RAG Retrieval (Retrieval-Augmented Generation)

A technique where an AI agent retrieves relevant information from external sources (like a product catalog) and uses it to generate responses or make decisions. RAG combines retrieval and generation.

Example: An agent retrieves BuyWhere search results for "red sneakers under $80", then uses those product details to generate a recommendation response with actual prices and retailer options.

Related Resources