DoorDash Brings AI Food Ordering to Apple Messages, Slack and Agent Workflows

DoorDash is testing an AI agent that lets customers order food through Apple Messages, turning a familiar text thread into a restaurant search, recommendation and checkout channel. Users can type requests such as “order my usual,” ask for a specific dish or seek a local recommendation. The system searches nearby restaurants, builds a suggested cart and can send photos of recommended food, according to the company.

The feature extends DoorDash’s in-app Ask DoorDash chatbot, which can interpret more open-ended requests such as recipes or photos. The text interface is designed to remove the need to open the DoorDash app for repeat orders and simple discoveries. Bloomberg described the launch as part of DoorDash’s broader effort to place its marketplace inside personal AI assistants, while TechCrunch identified Uber Eats and Grubhub as competitive pressure behind the move.

The agent is built around intent rather than rigid menu navigation. A customer can describe a craving, request a recurring meal or ask for options nearby, and DoorDash handles restaurant discovery before presenting a cart for approval. Group ordering is also part of the announced capability, although the available product details do not spell out every step of that workflow.

DoorDash’s approach keeps recommendations and ordering inside Apple Messages, where customers already communicate with friends and family. That matters because the service competes for attention with other conversational ordering tools, including Instinct-style agents and Square’s ChatGPT and Claude integrations. AOL characterized DoorDash’s feature as a beta, while coverage from The Verge also placed it alongside the company’s other agentic delivery experiments, including drone deliveries and retail returns.

DoorDash is extending agentic ordering beyond individual customers through DoorDash MCP, a Model Context Protocol offering for corporate ordering. Companies can use the system to automate office restaurant orders and supply restocks through workplace software, including Slack, according to Adweek. That creates a separate path into DoorDash’s marketplace, where a request can begin in a team conversation instead of a consumer app.

The workplace push gives DoorDash a way to serve recurring, multi-person transactions with software rather than a single customer. It also places the company in a race to make its menus, prices, availability and checkout functions readable by outside assistants. Square’s restaurant integrations show the same direction from the merchant side: conversational tools can expose live menus and modifiers, route orders to existing point-of-sale systems and label them as an AI channel.

DoorDash is also opening programmatic access through dd-cli, a command-line interface that allows AI agents to search restaurants, compare deals and complete real orders. Finance.yahoo reported that the tool can process real payments without requiring manual input at checkout. That shifts DoorDash’s AI strategy from a customer-facing chatbot toward infrastructure that other software agents can call.

The commercial upside is broader distribution across assistants, enterprise workflows and automated purchasing systems. The operational burden rises with it. Agents must handle substitutions, dietary requirements, payment authorization, delivery addresses and order errors accurately, while restaurants need menus that software can interpret. DoorDash’s earlier voice-ordering work for restaurants and its use of Replicant to automate more than 350,000 support calls per day show the company has been building automation across both sides of the marketplace.

DoorDash’s launch places the company in competition with delivery platforms and AI assistants over who controls the customer’s first interaction. Uber Eats and Grubhub remain direct rivals in food delivery, while Instinct and Square represent adjacent approaches that connect conversational software with transactions. The key contest is moving from app features to access inside the places where people already type requests.

That shift gives DoorDash a reason to expose its marketplace through Apple Messages, Slack and command-line tools at the same time. It also creates a new risk: when an assistant selects a restaurant, the platform’s ranking, pricing and recommendations become part of an automated decision rather than an explicit browsing session. DoorDash’s marketplace gains more entry points, but each one requires clear controls around consent, payments and the final cart.

DoorDash’s immediate next step is testing and refining the Apple Messages experience before broader adoption. Customers will judge the feature on whether “order my usual” reliably identifies the right meal, whether recommendations reflect real availability and whether the convenience beats opening the app. Group ordering and image-based suggestions will also test how well the agent handles ambiguous, multi-person requests.

Restaurants face a parallel adjustment as AI agents become another ordering channel. Accurate menus, modifiers, prices, hours and inventory will matter more because software needs structured information to make reliable purchases. DoorDash’s support automation and corporate tools indicate a larger effort to connect discovery, checkout and fulfillment, while Square’s integrations show that rival platforms are building similar paths. The winning service will be the one that makes automated ordering dependable without making customers feel they’ve lost control.