DoorDash AI agent ordering arrived on Wednesday 30 September 2026, when the delivery company said customers could now order food simply by sending a text message inside Apple Messages. Type “order my usual” and the agent knows you mean your Friday night takeaway, with the extra sauce and without the coriander. Ask for a dish or a recommendation and it searches local restaurants, suggests a basket, texts you photos of the food and checks out without you ever opening the app.
The DoorDash AI agent is in beta, and US customers can join a waitlist at doordash.com/text. It builds on Ask DoorDash, the in-app assistant launched in June, and arrives with two new grocery features and a drone delivery test in Northern California.
This article explains how the DoorDash AI agent works, what DoorDash’s own numbers say about AI ordering, how it compares with rival delivery assistants, why a text-based agent is a strategic move rather than a gimmick, and the questions customers should ask before handing over their usual order. For the earlier wave of delivery assistants, see our reports on Ask Shipt and Instacart’s Clementine.
Table of contents
- How the DoorDash AI Agent Works
- A Day With the DoorDash AI Agent: Three Scenarios
- What DoorDash’s Numbers Say About AI Ordering
- New Grocery Features Alongside the DoorDash AI Agent
- The DoorDash AI Agent Against Rival Delivery Assistants
- Why DoorDash Built a Text Agent Now
- Questions to Ask Before Texting the DoorDash AI Agent
- What the DoorDash AI Agent Means for UK Customers
- DoorDash AI Agent FAQ
- References and Further Reading
How the DoorDash AI Agent Works
DoorDash’s announcement describes the service as ordering “as simple as sending a text”. There is no app to open and no menu to browse. You message DoorDash the way you would message a friend.
Text what you want
You can ask for a specific dish, request a local recommendation based on your mood, or reorder a regular family meal. According to DoorDash, the agent then “searches local spots, scans menus, and suggests a cart based on what you’re in the mood for”. It even sends photos of the dishes it recommends, taken from the restaurant’s page on DoorDash.
It knows your usual
The headline feature is memory. DoorDash says that if you text “order my usual”, the agent “knows you mean your Friday night order”, including modifiers such as skipping coriander, adding extra sauce or never missing dessert. Bloomberg’s example is “order my usual protein bowl to the office”. The more you order, the more the DoorDash AI agent learns.
Group orders
Ordering for a group is the problem DoorDash chose to highlight. The agent “can handle mixed dietary preferences and different quantities in one order”, so one person does not have to collect everyone’s requests and rebuild them in the app.
Checkout in the thread
You confirm the order inside the text conversation and the agent handles checkout. Bloomberg reports that the service matches your phone number to your DoorDash profile, which is how it knows your order history, preferences and stored payment method.
What it runs on
DoorDash has not said which AI model powers the service or whether it built the model itself. What the announcement describes is a familiar stack: natural language processing to turn a casual message into a structured order, a recommendation system that already knows each customer’s history, and DoorDash’s existing checkout. The DoorDash AI agent is, in effect, a new front door to systems DoorDash has run for years, which is why it can launch with features such as stored payments and order history from day one.
It can message you first
One feature goes further than a normal chatbot. DoorDash says that, if you want, the agent “can message you proactively to ask if you want your usual coffee order based on your daily habits”. That turns the DoorDash AI agent from a tool you summon into one that prompts you.
A Day With the DoorDash AI Agent: Three Scenarios
DoorDash’s announcement lists features, but the clearest way to understand the DoorDash AI agent is to follow the jobs it is designed for. These scenarios are built from the capabilities DoorDash describes; the beta itself may behave differently.
The office lunch
A team of six wants lunch: two vegetarians, one person avoiding gluten, and someone who only wants a salad. Normally one person collects requests in a chat, opens the app and builds a group order by hand. With the DoorDash AI agent, that person texts the whole brief in one message. DoorDash says the agent handles “mixed dietary preferences and different quantities in one order”, so it should return one basket for the organiser to check and confirm.
The Friday usual
A family orders the same takeaway most Fridays. Instead of opening the app, finding the restaurant and re-adding each modifier, someone texts “order my usual”. The DoorDash AI agent recognises the regular order, including details such as no coriander and extra sauce, and asks for confirmation in the thread. This is the use case DoorDash leads with, because repeat orders are where convenience turns into habit.
The weekly shop
For groceries, the same logic applies to restocks. DoorDash says more than 30% of grocery orders through Ask DoorDash are routine restocks. A customer can ask for the usual weekly shop, then accept or reject the cheaper swaps DoorDash finds on the shelf. The agent does the repetitive part; the customer keeps the final say on substitutions.
What DoorDash's Numbers Say About AI Ordering
DoorDash backed the launch with figures from Ask DoorDash, its in-app assistant. They come from DoorDash’s own data and footnotes, not an independent audit, but they show why the company is pushing further.
| Claim | Figure | Data period in DoorDash’s footnote |
|---|---|---|
| Items available to an average US customer | More than 800,000 | Sample of 10,000 US consumers, June 2026 |
| New restaurants discovered via Ask DoorDash | More than 40,000 | June to August 2026 |
| Ask DoorDash restaurant orders to a place never tried before | Nearly half | June to August 2026 |
| Ask DoorDash grocery orders that are routine restocks | Over 30% | Mid August to mid September 2026 |
| Speed of building a grocery basket with Ask | 5 times faster | June 2026 |
| Basket value of Ask-built grocery orders | Nearly 50% higher | August to early September 2026 |
| Unique items in Ask-built grocery orders | About 60% more | August to early September 2026 |
Bigger baskets
The most commercially important figures are the basket ones. DoorDash says grocery orders built with Ask “average nearly 50% higher basket value and about 60% more unique items than the same consumers’ traditional orders”. For a delivery business, a bigger basket per trip is the difference between a profitable order and a loss-making one.
Ask-built grocery orders compared with the same customers’ usual orders, % higher (DoorDash figures, August to early September 2026)
Discovery
DoorDash says customers have used Ask DoorDash to find more than 40,000 new restaurants in its first three months, and that nearly half of Ask DoorDash restaurant orders go to places the customer has never tried. That is a useful point for restaurants: an AI that recommends can redistribute orders, for better or worse.
How to read these numbers
There is a caveat. The people who use a new AI assistant are early adopters, and early adopters may already order more. Comparing “the same consumers” removes some of that bias, but not the effect of novelty. The figures show what DoorDash wants to scale, rather than proving the DoorDash AI agent will produce the same results for everyone.
New Grocery Features Alongside the DoorDash AI Agent
The text launch came with two changes inside the DoorDash app, both aimed at weekly grocery shopping.
Restock with a Tap
Because more than 30% of grocery orders through Ask DoorDash are routine restocks, DoorDash is adding a prompt on your preferred grocer’s page to reorder your usual items. One tap builds the basket, and you can remove anything you do not need before checkout.
Suggested Swaps
When you build a grocery basket by asking DoorDash, it now “scans that store’s shelves in real time to find cheaper price swaps for the same or equivalent products”. It totals the potential savings before checkout, and you can accept or decline each swap.
Drone delivery tests
TechCrunch adds that DoorDash will begin testing delivery drones with selected restaurants in Northern California. The two launches fit together: an agent that removes the effort of ordering, and a delivery method that could cut the cost of the last mile.
The DoorDash AI Agent Against Rival Delivery Assistants
Every large US delivery platform now has an AI shopping assistant. The difference with the DoorDash AI agent is the channel: it lives in a messaging app rather than inside DoorDash’s own.
| Assistant | Launched | Where it works | Main job |
|---|---|---|---|
| Uber Eats Cart Assistant | 11 February 2026 | Inside the Uber Eats app | Building grocery baskets from text or photos |
| Ask DoorDash | 11 June 2026 | Inside the DoorDash app | Food, groceries and reservations by prompt or photo |
| Instacart Clementine | 9 September 2026 | Inside the Instacart app | AI grocery shopping assistant |
| Ask Shipt | 9 September 2026 | Inside the Shipt app | AI shopping assistant |
| DoorDash text agent | 30 September 2026, waitlist | Apple Messages | Ordering and checkout by text, including group orders |
The gap keeps shrinking
Measured from Uber Eats’ launch, Ask DoorDash came 120 days later, Clementine and Ask Shipt 210 days later, and the DoorDash AI agent 231 days later. The interval between DoorDash’s own two launches was 111 days. Features that once differentiated a delivery app now spread across the industry within a season.
Days after Uber Eats Cart Assistant (11 February 2026), by launch (our calendar arithmetic)
Why messaging is different
An assistant inside an app still requires you to open that app, while the DoorDash AI agent does not. A text thread is already open on most phones, sits beside messages from friends and family, and can send you a prompt. That is why the text channel matters more than any single feature: it moves DoorDash into the place where people already spend their attention.
Apple’s role
Choosing Apple Messages gives DoorDash instant reach on iPhones, but it also makes Apple the landlord. Apple controls how businesses can message customers, what those conversations can contain and which devices they appear on, and Android users are left out entirely at launch. If the DoorDash AI agent proves popular, DoorDash will want to reduce that dependence, either by adding other messaging channels or by making sure the same agent works inside its own app.
Why DoorDash Built a Text Agent Now
TechCrunch frames the launch as a way for DoorDash “to gain an edge over rivals Uber Eats and Grubhub”. There is a bigger reason too, and it is about who owns the customer.
General agents want to do the ordering
A new class of personal AI agents promises to handle errands for you. Instinct, which recently raised $1 billion at a $10 billion valuation, books travel, makes reservations and runs errands, and Yahoo framed DoorDash’s launch as arriving “as AI agents like Instinct are taking off”. Meta’s Muse and OpenAI’s new Dots agent aim at the same everyday tasks. DoorDash is also available as an app inside ChatGPT.
The risk of becoming a back end
If people tell a general assistant to “get dinner”, that assistant chooses the platform, and the delivery app becomes a supplier hidden behind it. That threatens the things that make delivery apps profitable: the customer relationship, the data on what people order, and the ability to sell promoted placements to restaurants. A DoorDash AI agent that people text directly keeps all three with DoorDash.
Agents that call on your behalf
The general agents are getting more capable. As we reported in our story on Instinct and Muse adding phone calls, they can already ring businesses for you. A delivery platform that does not offer its own agent invites those general agents to sit between it and its customers.
The economics of a bigger basket
The basket figures explain why an agent is worth building even before it is widely used. Delivery is expensive per trip: a courier, a pickup and a drop-off cost roughly the same whether the bag holds one item or ten. If AI-built grocery orders really are nearly 50% larger, each trip carries more revenue against the same delivery cost. The DoorDash AI agent extends that logic from the app to the text thread, where repeat orders can happen with almost no friction at all.
Andy Fang’s pitch
DoorDash co-founder Andy Fang put the ambition plainly: “We built a personal shopper that helps you discover and compare thousands of items while identifying ways for you to save on DoorDash.” He added: “Longer term, we’ll continue to find ways to make DoorDash even more useful before you need to ask.” That last phrase is the strategy: an agent that anticipates demand rather than waiting for an order.
Questions to Ask Before Texting the DoorDash AI Agent
Convenience brings trade-offs. DoorDash has not yet published detailed terms for the DoorDash AI agent, nor documentation for the beta, so these are the questions worth asking as it rolls out.
What happens if it gets the order wrong?
An agent that interprets “my usual” can misread it, especially if your usual changes. Check the basket summary before you confirm, and find out how refunds work for an order the agent built rather than one you assembled yourself.
Who can order on your phone?
Because the service matches your phone number to your account and checks out with your stored payment method, anyone with access to your unlocked phone and your messages could potentially place an order. Treat the text thread like a payment app and use your phone’s lock screen.
How much does it learn?
The agent learns your habits, including the times you usually order. Proactive messages are described as optional (“if you want”). Check how to turn them off and what DoorDash does with the history it uses to make suggestions.
Is the recommendation neutral?
When an AI suggests a restaurant or a cheaper swap, it is making a choice for you. DoorDash already sells advertising to merchants. Customers should watch whether promoted restaurants appear more often in the agent’s suggestions, and regulators may ask the same question.
Is it easier for everyone?
Texting can be easier than navigating an app for some customers, including people who rely on voice dictation or screen readers, and those who find busy menus hard to use. That could make the DoorDash AI agent genuinely useful beyond convenience. The flip side is that a text thread shows less information at once, so customers who want to compare prices, fees and delivery times side by side may still prefer the full app.
What about restaurants?
For restaurants, an agent that sends nearly half of its orders to places the customer has never tried is an opportunity and a risk. Visibility now depends partly on how an AI reads your menu, photos and reviews, so accurate item descriptions and good photos matter more than ever.
What the DoorDash AI Agent Means for UK Customers
The beta and waitlist are US only, and DoorDash has not announced dates for other countries. But the launch is relevant in the UK for two reasons.
DoorDash now owns Deliveroo
DoorDash bought UK-based Deliveroo in 2025, so its product choices now shape one of Britain’s largest delivery apps. Features that work in the US, such as Ask DoorDash and text ordering, are natural candidates to arrive in the UK, although nothing has been announced.
The messaging question
The US service runs in Apple Messages, and iPhone owners make up a large share of UK smartphone users too. Many people in the UK, though, do most of their messaging on WhatsApp. Any UK version of the DoorDash AI agent would need to decide which channel to use, and that choice would affect who can use it.
What UK restaurants can do now
UK restaurants on Deliveroo or DoorDash do not need to wait for a launch date to prepare. AI ordering depends on clean data: accurate dish names, complete allergen and dietary information, sensible modifiers and good photos. An agent that is asked for “something vegetarian and filling” can only recommend a dish it can recognise as both. Restaurants that tidy their menus now will be easier for any AI assistant to recommend, whether it is the DoorDash AI agent or a general assistant ordering on a customer’s behalf.
Rules on AI and consumers
A UK launch would also sit under consumer protection and data protection law. An agent that remembers habits, sends unprompted offers and recommends merchants would need clear consent for marketing messages and clear terms on how recommendations are ranked.
DoorDash AI Agent FAQ
What is the DoorDash AI agent?
A beta service, announced on 30 September 2026, that lets DoorDash customers order food and groceries by texting DoorDash inside Apple Messages. It builds a basket, sends photos and handles checkout.
How do I get it?
US customers can join the waitlist at doordash.com/text. DoorDash has not given a date for wider availability.
Does it work on Android or WhatsApp?
DoorDash’s launch describes Apple Messages only. It has not announced support for other messaging apps.
Can it order for a group?
Yes. DoorDash says it can handle mixed dietary preferences and different quantities in one order.
Does the DoorDash AI agent cost extra?
DoorDash’s announcement does not mention a separate fee for using the DoorDash AI agent. Orders placed by text go through the same checkout as orders in the app.
Is it available in the UK?
Not yet. The beta is US only, although DoorDash now owns Deliveroo in the UK.
References and Further Reading
DoorDash launches an AI agent you can text to order food (TechCrunch)
Ordering is now as simple as sending a text (DoorDash newsroom)
DoorDash unveils text-to-order AI agent that works with Apple’s iMessage (Bloomberg)
Ask DoorDash (DoorDash newsroom)
DoorDash’s new AI chatbot lets you order with prompts and photos (TechCrunch)
Uber Eats launches AI assistant to help with grocery cart creation (TechCrunch)
DoorDash now has a text-to-order option as AI agents like Instinct are taking off (Yahoo)
DoorDash unveils new text messaging AI assistant for ordering food (The Mac Observer)
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
