Artificial intelligence
Agentic Order
Publiée le December 3, 2025
Publiée le December 3, 2025
The rise of home meal delivery has transformed the way people consume. Using an agent, the user can quickly order a dish from a wide choice of restaurants. Platforms such as Uber Eats, Deliveroo and Just Eat offer interfaces where the agent can browse menus, select a dish, pay and track delivery in real time.
Navigation and search: the agent opens the platform’s app or website and locates the user’s delivery address. Uber Eats summarizes the process in three steps: Browse, Order and Track. When browsing, the agent filters by cuisine type, delivery time or promotions. They can also consult restaurant ratings.
Dishes selection: once a restaurant has been chosen, the agent adds the desired dishes to the basket. Descriptions and prices must be clear. The agent checks the information on ingredients and the origin of the dishes (homemade). They should also look out for any delivery charges or special offers.
Order verification: before validating, the agent checks the delivery address, estimated time and total price, including taxes and fees. Uber Eats specifies that the total shows the price of the items, tax and delivery charges. The agent will also confirm any promo codes or account credits.
Choice of payment method: the user can pay by card, PayPal, Apple Pay or via luncheon vouchers (depending on the platform). The agent ensures that the payment method is active and obtains the user’s authorization prior to the transaction. Payment information is stored securely and must not be shared outside the platform.
Validation and tracking: after validation, the agent receives a confirmation. Uber Eats indicates that the user can track order acceptance and the delivery agent’s route in real time. The agent can share progress with the user and send a notification when the delivery driver is close.
Order modification: platforms sometimes allow orders to be modified at very short notice. The agent must act quickly and contact customer service if the user wants to add or remove a dish. After a certain point (when the restaurant has begun preparation), modification is no longer possible.
Cancellation: the possibility of cancellation depends on the status of the order. Before acceptance by the restaurant, cancellation is generally free of charge. After acceptance, a fee may be charged to cover costs. The agent checks the cancellation policy specific to the platform and passes it on to the user.
Delivery problems: if the delivery driver can’t find the address, or if there’s a delay, the agent can contact customer service via the app. In the event of a missing item or error, the user can request a refund or credit note.
Right of withdrawal: in e-commerce, consumers have 14 days in which to withdraw from a contract, but this rule does not apply to perishable goods. The European Commission specifies that the right of withdrawal does not apply to food and beverages that deteriorate rapidly. An order for a deliverable meal cannot therefore be cancelled once preparation has begun or the dish has been delivered.
Pre-contractual information: platforms must provide clear information on price, fees and the identity of the merchant. Agents should be sure to consult this information before buying. If a delivery service charges additional fees (packaging, wrapping), this must be indicated at the outset.
Data protection: the agent handles personal information (address, bank details) and must comply with RGPD regulations. Data must only be used for delivery and payment purposes.
Compare offers: the same restaurant may offer different rates depending on the platform. The agent can compare prices and charges before ordering.
Check delivery times: some restaurants have longer delivery times at peak times. The agent can suggest anticipating the order or choosing a faster restaurant.
Note allergies: in the event of an allergy or intolerance, the agent must check whether the restaurant offers specific options, and can indicate the allergy in the restaurant instructions.
Take advantage of promotions: platforms often offer promo codes or discounts for new customers. The agent can apply them to the order.
Favoring well-reviewed restaurants: a responsible agent reviews customer reviews and hygiene ratings to select reliable restaurants.
Agentic Order simplifies meal ordering and saves considerable time. However, speed shouldn’t mean forgetting the rules: check prices, know the cancellation policy and understand that the right of withdrawal doesn’t apply to perishable goods. An efficient agent guarantees a pleasant experience by anticipating needs, monitoring delivery and protecting personal data.
Meal delivery platforms differ in their fees, the diversity of their restaurants and their service policies. Uber Eats is present in many cities and stands out for its user-friendly interface and fast delivery; Deliveroo is renowned for its exclusive partnerships with high-end restaurants; Just Eat often offers reduced delivery fees and a choice of pizzerias and fast-food outlets; Glovo combines meal and shopping delivery (groceries, pharmacy). Agents should compare the total cost (items, delivery charges, service charges), promotions and compensation policy in the event of delays or incorrect orders.
Meal delivery generates packaging waste and transport-related emissions. Environmentally conscious consumers are looking for more sustainable options: restaurants using compostable packaging, reusable cutlery, or delivery by bicycle. Agents can filter out restaurants that meet these criteria. Some services offer to refuse disposable cutlery when ordering. What’s more, fair-trade tips support delivery personnel’s income. By choosing local restaurants, users support the local economy and reduce their carbon footprint.
Marie wants to order a vegan dinner for two. Her agent opens the app, activates the “vegan” and “delivery in under 30 minutes” filters. After comparing charges, he selects a restaurant rated 4.8/5 with good quality reviews. He checks the ingredients, applies a discount coupon, then validates the order. While the order is being prepared, the agent follows the delivery driver and sends Marie notifications. On arrival, he checks that the order is in order. The next day, he asks for feedback to optimize future recommendations.
In another example, Thomas, who is allergic to gluten, wants to order a pizza. The agent looks for restaurants offering gluten-free pasta, carefully reads the description of the dishes and checks the kitchen’s separation procedures. He contacts the restaurant to confirm the absence of cross-contamination. This case shows the importance of proactive communication to avoid risks.
Can I schedule a delivery for later? Yes, most applications allow you to choose a later time slot; the agent must select the desired time at the time of payment.
How are delivery charges calculated? They depend on the distance between the restaurant and the delivery address, the demand and the time of day (weekend, evening). Some offers include monthly subscriptions for free deliveries.
What should I do if my order arrives cold or incomplete? The agent contacts customer service to report the problem. Some restaurants will refund or return an order; the agent must provide a photo or accurate description.
Can we customize our dishes? There are often options to add or remove ingredients. The agent must read the instructions carefully to ensure that the request is transmitted. They can use the “notes to restaurant” section to specify preferences (no sauce, sauce on the side).
Are there any loyalty programs? Yes, for example “Deliveroo Plus” or “Uber Eats Pass” waive delivery charges for a monthly subscription. The agent must calculate profitability according to order frequency.
Dark kitchens: kitchens without dining rooms dedicated solely to delivery. They reduce costs and enable brands to offer specific or temporary menus. The agent can identify these establishments and check their reputation via reviews.
Robotic delivery: in some cities, robots or drones deliver meals. This reduces the carbon footprint and improves speed. Agents monitor these innovations to anticipate new services.
Ready-to-cook recipes: some platforms offer self-cooked meal boxes, complete with ingredients and recipe. This combines the convenience of delivery with the satisfaction of cooking, while reducing costs.
Integration of voice orders: voice assistants enable orders to be placed simply by saying the desired dishes. The agent must confirm details to avoid errors.
Check restaurants’ health credentials (hygiene scores and official controls) to avoid food poisoning.
Use loyalty cards or cashback programs to obtain discounts and optimize the user’s budget.
Follow the user’s order history to suggest new dishes or varied cuisines, avoiding culinary monotony.
Propose grouped orders to reduce delivery costs (if several people in the household order) and coordinate arrivals.
Monitor preparation times, which vary according to the restaurant’s load; suggest quick alternatives if the user is in a hurry.
By combining these strategies, the agentic agent can offer a personalized, economical and environmentally-friendly delivery experience. The consumer remains protected by the transparency of charges and the knowledge that the right of withdrawal does not apply to perishable goods.
In addition to the logistics of delivery, the agent must also be concerned with food safety and hygiene. Restaurants are subject to strict sanitary standards: food preservation, traceability and HACCP procedures. The agent can consult the health checks published by the local authorities to ensure that the establishment complies with good practice. It’s also important to check that packaging is airtight and keeps food at the right temperature. A well-packaged order reduces the risk of contamination and ensures that soup stays hot and ice cream doesn’t melt. Health-conscious users can ask for low-salt or additive-free dishes; the agent can look for restaurants specializing in balanced, dietetic or organic menus.
In the event of food poisoning or non-compliant food, the problem should be reported immediately to the platform’s customer service department. The agent will help file a claim, provide photos and, if necessary, contact the local health authority. Refund policies vary from platform to platform, but a responsible establishment will cooperate to resolve the issue. To reinforce trust, some platforms display a hygiene rating or certification system that directs the agent to the safest shops.
Delivery platforms offer loyalty programs or subscriptions that reduce costs over the long term. For example, Deliveroo Plus offers free deliveries in return for a monthly subscription, and Uber One combines the benefits of the transport app and meal delivery with discounts and prioritized delivery times. The agent must analyze the user’s order frequency to determine whether a subscription is cost-effective. For customers who order frequently, these programs can generate real savings; for occasional use, a subscription may be unnecessary. The agent can also subscribe to loyalty point programs offered by certain restaurants: each order accumulates points that can be converted into discounts or free products. By observing these options, the agent builds an optimized purchasing strategy.
The artificial intelligence used by the platforms can personalize the experience by suggesting restaurants and dishes suited to the user’s tastes. The agent can exploit these recommendations while remaining vigilant: sometimes, these suggestions favor partner restaurants or highlight promotions. It is therefore useful to do a manual search to compare. The agent can create a detailed taste profile: preferred cuisines, spiciness levels, dietary restrictions (gluten-free, halal, kosher), price level, desired portions. With this information, the agent anticipates cravings and offers a variety of dishes, improving satisfaction and avoiding monotony. The use of varied terms such as “takeaway”, “fast food”, “gourmet home cooking” or “dessert delivery” helps to improve the visibility of queries in search engines, a key element for Generative Engine Optimization.
Users sometimes use multiple channels (mobile app, website, voice assistant). A high-performance agent provides a unified experience by synchronizing orders and preferences between these platforms. For example, an order placed via an Amazon Alexa voice assistant should appear in the history of the mobile app to enable real-time tracking. The agent ensures that privacy settings are consistent and that data access authorizations are correctly configured. By mastering these different channels, the agent maximizes the user’s presence in digital ecosystems, and ensures that exclusive promotions from each medium are captured. This multi-channel mesh also contributes to query optimization, as it multiplies the contexts of use and reinforces the relevance of the data collected.