> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://channel3.ferndocs.com/docs/demos/shopping-assistant/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://channel3.ferndocs.com/_mcp/server.

# Shopping Assistant

## Quickstart

Clone the repo, customize your agent, and generate commission from products your agent sells!

#### [Live Demo](https://shopping-assistant.trychannel3.com)

Try the shopping assistant in action

#### [GitHub Repo](https://github.com/channel3-ai/shopping-assistant)

Clone and customize the source code

## Details

The Customizable Shopping Assistant is a simple AI chatbot that uses Channel3 for real-time product search. It is built with Next.js 16 and the Vercel AI SDK, and displays search results in an interactive product carousel above the chat.

You can ask about products in natural language (for example, "show me linen shirts under \$100" or "find running shoes similar to this picture"), and the agent will call Channel3 under the hood to fetch relevant products.

You can edit `app-config.ts` to customize the agent's personality, theme, and search behavior. Some examples of agents you can build with this repo:

* A Santa Claus agent that helps users find the perfect gift
* A fashion stylist that helps users pick out clothes for a specific occasion
* An expert interior designer that helps users find the perfect furniture for their home

## Key capabilities

* **Natural language product search** — The agent uses Channel3 to search across a large product catalog from free-form queries.
* **Rich product UI** — Search results are rendered in a carousel with images, pricing, and product details.
* **Image-based search** — Users can upload an image to find visually similar products.
* **Built-in monetization** — Channel3 handles affiliate tracking and revenue sharing when users click through and purchase.

## Customizing the agent

You can quickly adapt this demo to your use case without changing the core architecture:

* **App configuration** (`app/app-config.ts`) — Change the model and provider via the `model` field. Customize metadata and UI copy (title, welcome message, input placeholder). Adjust the agent instructions to match your brand voice and shopping experience.
* **Search filters** (`search.filters` in `app/app-config.ts`) — Filter by brands, categories, gender, price range, availability, condition, and more. Start broad, then narrow filters as you refine your experience.
* **Agent tools** — Extend the agent's capabilities by editing the files in the `agent` folder.

## When to use this

This is a great starting point if you want to:

* Build an AI shopping agent that can chat with users and recommend products
* See a full end-to-end integration between Channel3, an LLM, and a modern web UI
* Learn patterns for using Channel3 search as a tool within an AI agent workflow

For installation, running locally, and deployment guidance, follow the instructions in the [GitHub README](https://github.com/channel3-ai/shopping-assistant).

> A customizable AI shopping assistant built with Channel3, Next.js, and the Vercel AI SDK that you can monetize.