How It Works — CartIQ by NULogic

How It Works

From zero to production agent in 15 minutes

No AI expertise needed. Connect your platform, add your brand config, deploy to your cloud.

017 platforms supported

Connect Your Platform

Point the adapter at your SFCC, commercetools, Shopware, or Shopify instance. Add your API credentials. The agent gains full access to your live catalog, inventory, pricing, and cart.

.env

RETAILER=my_store
CTP_PROJECT_KEY=my-project
CTP_CLIENT_ID=xxxxx
CTP_CLIENT_SECRET=xxxxx

02~50 lines of code

Configure Your Brand

Write ~50 lines of Python that define your brand identity — colors, logo, greeting, selling style, payment methods. That's your entire brand configuration.

brand.py

@register_retailer
class MyStoreConfig(RetailerConfig):
    name = "My Store"
    brand_color = "#0057A0"
    greeting = "Hi! How can I help?"
    selling_style = "expert_advisor"
    platform = SFCCStore

03< 15 minutes

Deploy to Your Cloud

Run Docker on your GCP, AWS, or Azure account. Your agent is live — fully branded, connected to your catalog, serving your shoppers. Your data never leaves your infrastructure.

terminal

docker build -t my-agent .
docker run \
  --env-file .env \
  -p 10999:10999 \
  my-agent
# ✓ Agent live at https://agent.mystore.com

Architecture

3 layers. Only 1 you write.

NULogic uses a 3-tier architecture. You only write the brand config layer — everything else is pre-built.

Your Brand Config

~50 lines of Python. Defines your identity, personality, payment methods, and selling style. This is the only thing you write.

Brand colors & logoSelling style (luxury, expert, friendly)Payment methods to enableCustom greeting

Platform Adapter

Handles all communication with your commerce platform. Pre-built for SFCC, commercetools, Shopware, and Shopify. You don't write this.

Product search & filteringCart managementCustomer auth (OTP, social)Order placement

NULogic Core Agent

The AI engine — built on Google ADK. Understands natural language, orchestrates tool calls, manages conversations, tracks analytics.

Gemini 2.5 Flash (default)52 shopping toolsA2A/REST/MCP protocolsBuilt-in analytics

↑ What you see: 100% your brand · What you build: ~50 lines of config

Platform Adapters

Your platform. Our adapter.

Pre-built adapters handle all the commerce platform complexity. You connect your credentials and go.

SFCC

Ready

SW6

Ready

Shopify

Ready

Magento

Ready

VTEX

Ready

Protocols

One agent. Every channel.

The same CartIQ logic serves REST, A2A/UCP, and MCP clients simultaneously — including the external AI agents coming to your store next.

🛡️

Why these protocols matter beyond today

MCP and UCP/A2A are the standards external AI agents — Google AI Mode, ChatGPT Shopping, Perplexity — will use to interact with commerce systems. CartIQ implements both from day one. When external agents arrive at your store, they talk to CartIQ first: your guardrails, your catalog logic, your policies. Not raw platform APIs.

REST / JSON

Default

Simple HTTP endpoints — compatible with any client. What the built-in React frontend uses.

Your own frontend
GPT Actions
Any HTTP client
Mobile apps

A2A / UCP

Google AI Mode

Agent-to-Agent protocol. Powers Google AI Mode integration and Vertex AI agent connections.

Google AI Mode
Vertex AI
Agent orchestration
Enterprise AI platforms

MCP

AI Dev Tools

Model Context Protocol — lets AI coding assistants and Claude use your shopping agent as a tool.

Claude Desktop
Cursor IDE
Any MCP client
AI automation

Want to test the API directly?

Try live requests against our demo agents — REST, A2A agent card, or MCP tool discovery.

Open API Playground