DoorDash launches a beta CLI tool for ordering via AI agents

DoorDash has introduced a limited beta of DoorDash CLI, a command-line tool that lets developers place DoorDash orders directly through an AI agent. Called “dd-cli,” the tool can search stores, find deals, and complete checkout, according to the company. It is currently open to U.S. and Canadian macOS developers through a waitlist, DoorDash co-founder and CTO Andy Fang said in a post on X.

The tool effectively exposes DoorDash’s ordering platform to AI agents, letting developers build their own software around it — from custom apps for ordering food and groceries to tools for surfacing local deals — or combine it with other building blocks in an agentic workflow.

DoorDash CLI extends a pattern the company has already established elsewhere. DoorDash has previously experimented with ordering through iMessage, built its own AI chatbot called “Ask DoorDash,” and exposed its service to AI chatbots including OpenAI’s ChatGPT and Anthropic’s Claude. The sign-up form for the CLI beta asks developers what they would build if granted access.

A demo video attached to the announcement showed the agent working through a lengthy sequence of steps — reading Slack messages, recalling past interactions, parsing JSON, inspecting menu structures, running Python scripts, recovering from errors, and calculating totals — just to complete an order for three salads, illustrating both the tool’s flexibility and the extra complexity agentic workflows can add to a routine task.

Why it matters

DoorDash’s CLI is a concrete example of agentic commerce: rather than building a single chat interface, DoorDash is turning its ordering platform into infrastructure developers can build on directly, a model other consumer platforms are likely to test as AI agents become more capable of executing transactions.