DoorDash has launched a command-line tool that lets developers and AI agents order food. It is called dd-cli, and it is a glimpse of how commerce will work when AI agents are the customers.
The tool is in limited beta, but the direction is clear. Software is increasingly being designed for AI agents rather than just humans. DoorDash built an interface that an AI agent can use to search stores, build a cart, and place an order — all from a terminal.
This is not just a novelty for developers who want to order lunch without opening a browser. It represents a fundamental shift in how companies should think about their customer-facing technology.
What AI agents need from your business
The key insight behind dd-cli is that AI agents cannot use graphical user interfaces. They cannot click buttons, scroll through menus, or read images. They need structured, text-based interfaces they can parse programmatically.
DoorDash is not alone in recognising this. Companies across industries are building APIs and command-line tools specifically for AI agent consumption. The message is simple: if your business does not have a way for software to interact with your services directly, AI agents cannot use you.
For a small business, this might mean ensuring your website has a clear API, or that your online ordering system can accept requests from a bot as well as a person. Increasingly, the customer that shows up at your digital door might not be human.
The agent commerce chain
Imagine how this plays out in practice. A consumer tells their AI assistant: “Find me a good pizza place within 5 kilometres that has vegetarian options and can deliver in under 40 minutes.” The assistant searches available services, checks menus and delivery times, and returns a recommendation.
The user approves. The assistant places the order, pays, and sends a notification when the food is on its way. The entire transaction happens without the user ever opening a delivery app.
This is not science fiction. DoorDash is building the infrastructure for it right now. Visa is preparing for it with its Agentic Ready programme. Payment processors are developing agent-specific tokenisation and fraud monitoring to handle transactions initiated by software rather than people.
What this means for non-tech businesses
You might think this only matters for food delivery and tech companies. But the trend affects every business that sells products or services online.
If your business has an e-commerce site, AI agents will start shopping there. If your booking system requires a human to click through a calendar, agents will struggle. If your pricing requires visual parsing of a PDF menu, agents cannot read it.
The businesses that will benefit are the ones that make their services machine-readable now. That means structured data, clear APIs, and text-based ordering that does not rely on graphics or JavaScript-heavy interfaces.
You do not need to build a command-line tool like DoorDash. But you should ask: could an AI agent complete a transaction on your site without human intervention? If the answer is no, you are already falling behind.
The practical takeaway
DoorDash’s dd-cli is a small sign of a big change. AI agents are becoming real economic actors. They will search for products, compare prices, and place orders on behalf of humans. Businesses that prepare for this now will capture that new wave of demand.
The simplest first step: audit your customer-facing systems for machine readability. Then make sure your website works without JavaScript, your product catalogue has structured data, and your checkout can accept API requests. That is all it takes to be ready for the agent economy.