DoorDash Tests iMessage AI Agent That Can Turn a Text Into a Food Order

Late-night gaming sessions have long had a familiar companion: the debate over what to eat. It can be a bigger decision than choosing the next multiplayer match, especially when a party wants to keep playing rather than cycle through restaurant listings, compare delivery times, and negotiate toppings. DoorDash is looking to reduce that particular interruption with a new AI-powered ordering feature that lets customers send a text describing what they want.

The feature, currently available as a test that users can sign up for through DoorDash, connects people with an AI agent in iMessage. Rather than requiring customers to browse the usual restaurant menus in the DoorDash app, the agent is meant to interpret a message, search nearby options, and return recommendations informed by both the request and a customer's previous orders.

For now, the trial is focused on iOS. Android support is planned for the future, though DoorDash has not said when it will arrive. There is also no announced date for the feature to leave beta and become broadly available.

A text-first approach to ordering dinner

The central idea is straightforward. A customer texts the DoorDash AI agent with an order concept instead of manually navigating menus. That message could be broad, such as asking for something comforting for dinner, or highly specific, such as requesting pizza for a group. The agent then considers restaurants in the user's area and suggests choices based on the text and past purchasing behavior.

For useful background on this topic, read European Commission Targets Major Game Companies Over Virtual Currency Practices.

DoorDash has used the example of someone asking for their normal Friday-night order. In that scenario, the agent is intended to recognize the customer's typical selection and any usual customizations. That could make the tool relevant for people whose food habits are especially predictable: the same pizza, the same side, the same drink, or the same post-session meal after a regular online game night.

Personalization is doing much of the work here. Restaurant discovery is useful when a person cannot decide what they want, but repeat-order recall may be the more practical part of the feature. Regular customers often have a saved pattern already, whether that means a modified burger order, a particular noodle dish, or a pizza built around a familiar set of toppings. The proposed iMessage flow is designed to surface those routines through conversation.

That matters for convenience-oriented apps because the question is not always whether users can place an order quickly. It is whether they have to think through the entire process at all. Someone may know they are hungry without knowing which restaurant to choose, and an AI agent offers to narrow that choice down from a simple prompt.

How it could fit into gaming routines

Food delivery is not game software, but it sits close to the social routines around games. Friends preparing for a lengthy co-op session, a weekend raid, a local tournament, or a watch party may be more interested in preserving momentum than browsing an app. A text-based assistant could potentially be useful in those moments if it understands requests framed around a group: food that travels well, repeat favorites, or an order similar to what the user placed the last time friends came over.

There is a natural appeal to handling this from the messaging app that many groups already use for plans. Instead of switching between a chat, a game lobby, and a delivery app, an iPhone user in the beta could message the agent with a request. The system is not presented as a gaming-specific service, and there is no indication that it connects to consoles, PC games, voice chat, or community platforms. Its relevance is simpler: it aims to make an ordinary break-time task take less attention.

Still, the feature's usefulness will depend on how reliably it deals with the details that matter most in a group order. "Our usual Friday meal" sounds convenient when it accurately includes the right modifications, quantities, and restaurant. It becomes less helpful if customers need to repeatedly correct assumptions or manually reconstruct every item. Food ordering is full of personal preferences, and recalling a previous order is not the same thing as understanding whether that order still suits the current situation.

For example, a player ordering alone may want their standard meal. A group of four may need an entirely different cart, and a casual text prompt can leave ambiguity around portions, dietary needs, side dishes, or who is contributing. DoorDash's test is built around recommendations and ordering assistance, but the real measure of its value will be whether the conversational process removes friction rather than adding a new layer of it.

Convenience versus the app's existing shortcuts

There is an obvious question facing the new system: how much faster is it than the DoorDash app already is? Delivery apps commonly place recent orders and favorites near the top of the interface. For a repeat purchase, opening the app, selecting a prior order, checking it, and submitting it may already be a short process.

In some cases, texting an AI could even involve more steps. The customer must phrase a request, wait for the agent's response, review the recommendation, confirm any changes, and approve the order. If the request is simple and the past order is readily available in the app, the conventional interface could remain the more direct option.

The potential advantage is less about shaving every possible tap from a repeat order and more about avoiding the browsing stage. A person who does not know what they want may prefer typing a natural-language request over filtering a long collection of restaurants. The AI agent can take a vague preference and offer candidates, ideally with enough awareness of the user's order history to make those recommendations feel relevant.

That distinction is important. An app interface excels when someone knows exactly what to buy and wants full control. A conversational agent is meant to help when a person wants a suggestion, has a habit they want repeated, or would rather describe the outcome than search for it. DoorDash is testing whether customers find that trade worthwhile.

Text ordering has older roots than the AI label

Although the tool is built around the current wave of AI agents, the basic notion of ordering food through messages is not new. Before smartphone apps became the default route for delivery, SMS-based ordering was one way businesses tried to make repeat purchases easier. Pizza chains in particular have experimented with low-effort ordering systems for years.

Domino's, for instance, introduced an emoji-based ordering option in 2015. That system was limited to one restaurant brand, while DoorDash's approach is intended to search a broader range of restaurants available in a user's location. The newer feature also emphasizes conversational recommendations and order-history personalization rather than one ultra-short command tied to a preconfigured favorite.

In that sense, DoorDash's iMessage experiment combines an old interaction model with newer software expectations. The message itself is familiar, but the agent is supposed to do more interpretation behind the scenes: determine what the user means, compare options nearby, remember established preferences, and guide the order toward completion.

iOS beta now, Android later

Anyone interested in trying the feature can register for the test through DoorDash's website, but the current limitation is significant. The AI ordering agent works through iMessage, so it is presently an iOS proposition. Android customers will need to continue using existing ordering methods until support is introduced.

DoorDash has said Android compatibility will come later, but it has not provided a timetable. That means the early test will primarily reveal how iPhone users respond to the text-driven experience, not whether the concept works equally well across the broader mobile market.

There are also unanswered questions that naturally accompany a beta of this kind. DoorDash has not detailed when the agent will be released beyond the testing phase. Its long-term value will likely hinge on the quality of restaurant suggestions, the accuracy of remembered modifications, how clearly it communicates order details before checkout, and how easily users can override an incorrect assumption.

An experiment in making an easy task feel easier

Ordering delivery is already one of the more streamlined tasks on a phone, which makes this an interesting test of how far AI agents can push convenience. It is not solving a problem that every customer will feel strongly. Many users may prefer the visibility and certainty of scrolling through menus themselves, particularly when they are trying somewhere new or carefully comparing prices and options.

But for people who repeatedly order the same meals, struggle to decide, or simply want to keep a gaming night moving, a text conversation may prove more natural than opening another app. The pitch is ultimately not that customers are incapable of ordering food on their own. It is that a service which remembers habits and offers a quick prompt-based route may make the decision feel less demanding.

Whether that becomes a standard part of delivery ordering will be decided after the beta has had time to show how well the agent handles real requests. For now, DoorDash is inviting iOS users to test the idea: send a message about dinner, let an AI assistant search the local options, and see whether the result is more convenient than the app screen it is trying to replace.