AI is rapidly changing commerce, and consumer brands and platforms are on the move to respond.
Claude for commerce
Win the sprint to holiday season this year
Claude helps retail, travel, telecom, and entertainment companies build and deploy consumer and merchant AI agents in your app, connected to your systems, in weeks.
How commerce companies use Claude in agents
8 live demos
Trusted by
Built to bring the best of Claude to commerce agents
Holiday revenue this year, not next
Speed up production with an easy setup. Fork the repo to choose your agent type, industry, and how to build. Go live before code freeze to capture this year's holiday traffic in weeks.
Bigger baskets on your storefront
Shoppers get help finding, comparing, and deciding inside your app, then check out without going to a marketplace or third-party assistant. Your relationships stay yours.
More GMV per seller, with the team you have
One merchant agent covers the ground that used to take a few operators: catching price drift, keeping your catalog current, or drafting promotions. Sellers grow revenue, not the team.
Enterprise safety on every token
Safety patterns running today's largest commerce agents are built in: pricing and availability accuracy, cart and refund authority limits, and escalation to a human.

Retailers running shopping agents on Claude have seen carts up to 35% larger and shoppers 60% more likely to complete a purchase.
Our best practices from commerce teams in one forkable repo: two agents, four vertical implementations, and a Claude Code plugin to wire it to your own catalog.
What businesses are saying
Build quickly for customers and merchants
The same architecture, tools, and setup with an intuitive presentation layer for customer-facing and internal agents. Pick the skills you need, tuned to your industry.
Consumer agents
Embed a conversational layer in your app to help customers move from intent to your cart.
Skills
- Catalog search and discovery
- Planning and shopping lists
- Cart actions and checkout
- Customer care
- Personalization and memory
Merchant agents
Run store operations through internal agents so merchants can question performance.
Skills
- Sales analytics and insights
- Catalog and inventory management
- Promotions
- Campaign management
- Personalization and memory
Choose how you build on Claude
Flexible ways to build your commerce agent, fitted to the way you work. The repository comes with reference implementations for building each agent using Claude’s frameworks.
Messages API
Direct model access for full control over your agent loop.
Claude Agent SDK
The same agent loop and tools that power Claude Code, in Python or TypeScript, for teams that want to build the harness themselves.
Claude Managed Agents
Build production agents with a composable suite of APIs, without managing infrastructure.
Claude Code plugin
Use the plugin to take the blueprint and your chosen reference implementation and start building in Claude Code. Available in the repo.
higher conversion for small businesses selling through sales agents
FAQ
About the offering
It's an open reference implementation, not a supported product. You fork the repository and own what you build from it. Anthropic doesn't maintain it or accept contributions, and there's no service level agreement on the reference code itself.
No. Claude is your intelligence layer, not a storefront or checkout. We're not interested in owning catalogs, supply chain, or the last mile, and there's no advertising or paid placement of Claude in any agent you build. Your relationships stays yours.
The four reference implementations are starting points. Retail, travel, telecom, and entertainment ship with the repository, and the agent pattern isn't limited to any of them.
No. The setup flow walks through where the agents run, which systems they connect to, and which workflows you want to change. Start with the skills that match your surface and leave the rest.
Trust and safety
The consumer agent is structurally prevented by the harness from inventing a price or a product, and manipulative upsell patterns are prohibited. Cart and refund authority have hard limits, and anything past them escalates to a person.
By default, it drafts and stages changes, then waits for your review. Nothing reaches your storefront until a person approves it, and it can only act across the backend systems you have chosen to connect. If you prefer to bypass this process, you could tweak the harness to automatically accept changes based on a policy you define.
The Claude Code plugin writes the tests for you. It builds a suite that runs against your catalog, prices, and rules. We didn’t include a generic test set in the repository because it wouldn't tell you much about your deployment.
Through GitHub private vulnerability reporting. The repository documents its threat model, its defense in depth layers, guidance for adversarial evaluation, and what adopters are responsible for. Demo backends are not production hardened.
Bring Claude to your commerce agent
Fork it, run it, and ask Claude Code to make it yours.