Skip to main content

​
Better Data platform concepts

​
Layered ecosystem model (canonical)

Better Data is described as five layers — not as a flat list of products:
LayerSurface
InfrastructureGateway Console, Hosted Gateway MCP
ProtocolCommerce Gateway, Registry MCP
ReferenceGateway Demo, Commerce Demo, Agent Starter
ControlLoop Engine
PlatformBetter Data
  • Infrastructure is where operators configure and host gateway/MCP surfaces.
  • Protocol is what agents and integrations speak (gateway execution, registry discovery).
  • Reference is implementation demos and accelerators — illustrative, not control planes.
  • Control is governed automation (loops, policy, audit).
  • Platform is the workspace shell that composes the layers after onboarding.
→ MCP surfaces · Platform architecture

​
Agent flow (C04 / C12)

Agents reuse the layered stack — they do not replace it or sit “beside” the gateway as a separate product. Canonical wording: agents use Registry to discover, Gateway to act, and Loop Engine to govern. 
Agent
  ↓
Registry MCP
  ↓
Commerce Gateway
  ↓
Loop Engine
  ↓
Demo / UI
Experience (demos, product UI) sits at the bottom; developers start from Build an agent. See Commerce Agent pattern.

​
Loops, actors, guards, and evidence

The core concepts in Better Data apply across all four platform layers (loop / gateway / SCM·DCM / Signal Tags — see Platform architecture):
  • Loops — governed workflows. Every operation is a loop.
  • Actors — human, automation, or AI agent. Every action is attributed.
  • Guards — policies enforced at runtime. Not in prompts.
  • Evidence — structured context attached to every decision.
  • Audit trail — immutable record of every loop transition.
The Commerce Gateway, SCM/DCM modules, and Signal Tags all participate in loops. The loop model is the unifying concept. → Platform Architecture
→ Loops

​
Commerce Gateway concepts: Backends, Tools, and Sessions

The Commerce Gateway is built on three main pillars: Backends, Tools, and Sessions. Understanding these concepts will help you design flexible and robust conversational commerce experiences.

​
1. Backends

Backends are where your data lives. They provide the gateway with access to your product catalog, shopping carts, and order management systems.
  • ProductBackend: Handles searching and retrieving product information.
  • CartBackend: Manages the lifecycle of a shopping cart.
  • OrderBackend: Facilitates order creation and status tracking.
Learn more about Backends ->

​
2. Tools (Capabilities)

Tools (often referred to as Capabilities in the marketing docs) are the specific actions an AI can perform. The gateway takes your backend logic and “adapts” it into a format the LLM can understand (like MCP for Claude or Function Calling for OpenAI). Common tools include:
  • search_products
  • get_product_details
  • add_to_cart
  • check_inventory
Learn more about Tools ->

​
3. Sessions

Sessions maintain the state of a conversation across multiple turns. They store things like the current cartId and user context.
  • Persistent: Sessions can be stored in Redis to survive server restarts.
  • Cross-Platform: A session started in Claude can be transferred to a web checkout or even another AI assistant.
Learn more about Sessions ->

​
Architecture Overview

The Commerce Gateway acts as the orchestration layer between various LLM providers and your commerce infrastructure.
This diagram is LLM ↔ gateway ↔ backends plumbing only. The canonical platform / layered ecosystem diagram for C8 is the layered ecosystem model above and Platform architecture.

​
Getting Started

If you’re ready to start building, head over to the Quickstart guide.