OrderJam Logo
    Home
    ← Back to OrderJam Journal
    Industry·7 min read

    An AI Took My Dinner Order Last Night. I Tipped It.

    Text-message ordering is having a quiet moment. After three weeks of using it at every restaurant that offered it, here’s what works, what doesn’t, and why your phone’s home screen is about to get a little less crowded.

    OE
    OrderJam Editorial
    Editorial team · April 22, 2026
    An AI Took My Dinner Order Last Night. I Tipped It.

    Photo by Patrick Tomasso on Unsplash

    The first time it happened I was standing in line at a deli on 9th Avenue. The woman in front of me was on the phone with a friend, and her friend was placing an order at the same restaurant via text message while we waited. By the time the friend got to the counter, her sandwich was already in the bag.

    I took out my phone and texted the number on the chalkboard. “Hi, can I see the menu?” A reply came back in about four seconds with the day’s specials and a polite question about whether I wanted pickup or delivery. It felt, weirdly, like talking to a host who was a little too caffeinated.

    The category nobody is naming yet

    What I was using is sometimes called “conversational commerce,” sometimes “AI ordering,” sometimes just “text the restaurant.” The branding is unsettled because the experience is still unsettled. Some restaurants route texts to a real employee. Some use a chatbot built on a large language model. Some use a hybrid — the AI handles 80% of the conversation and a human steps in when something goes sideways.

    The reason it’s spreading isn’t mysterious. The two existing channels for ordering food online are both broken in different ways. Delivery aggregators take 18–30% in commission and own the customer relationship. Native restaurant apps require a download, a login, and a slot on a screen that’s already full. SMS sits in between: it’s the most-used app on every phone, and it requires nothing.

    The most-used app on every phone is the one nobody calls an app.

    Three weeks, eleven restaurants, one observation

    I tried this at eleven places — a poke spot, two pizzerias, a Vietnamese place, a halal cart, a Mediterranean chain, a smoothie bar, a steakhouse, a sushi place, a wing joint, and a vegan café. Of the eleven, nine answered within ten seconds. The other two answered within a minute. None of them made me sign up for an account.

    The good parts were uniformly good. The friction is gone. There’s no app, no password, no “allow notifications” popup, no cart that resets when you close the tab. You text. The restaurant texts back. The thread sits in your phone next to your group chats, and next time you order from the same place you just type “the usual.”

    What worked

    • Speed. Median time from first message to confirmed order was under three minutes across the whole sample.
    • Memory. Six of the eleven remembered me on the second order. One remembered my partner’s allergy. None of them — to be clear — charged me for the privilege.
    • Politeness. Every system I tested was disarmingly polite. Two were funny. One was, I think, flirting with me. (It was the wing joint. I tipped well.)

    What didn’t

    Two of the bots got tripped up on modifiers. I asked one for “ranch on the side, no celery, extra blue cheese, and can you make the celery into carrots” — a sentence that would confuse a teenager — and it asked me to repeat the part about the carrots. Fair. Another one quoted me a delivery fee that turned out to be from the wrong courier, which the restaurant manager fixed with an apologetic phone call. None of the eleven flat-out failed. But three of them needed a small assist.

    There are also things SMS just isn’t good at. You can’t see the food. You can’t scroll a menu the way you can on a glossy app. If the menu has 200 items and you don’t know what you want, text-only is worse than a picture-driven UI. Most of the systems I tested handled this by texting back a link to a clean, image-rich web menu — SMS for ordering, web for browsing. The hybrid worked.

    The pros and cons, on one page

    If you’re a customer trying to decide whether to use these things, here’s the honest version:

    The case for

    • No app, no login, no friction.
    • Works on every phone made in the last twenty years.
    • Records itself. Your order history is a thread you can scroll back through.
    • Doesn’t spam you. (Well-built ones don’t. We’ll get to the badly built ones.)

    The case against

    • Bad implementations text you marketing at 7am. Avoid these.
    • The AI is only as good as the menu it’s trained on. Small, weird menus do better than big, sprawling ones.
    • If you want to browse rather than order, you still want a menu page.
    • Cross-restaurant search doesn’t exist yet — you have to know which place you’re texting.

    Why restaurants are doing this

    The restaurants I talked to had remarkably consistent reasons for adopting it. Three came up over and over:

    1. The phone never stops ringing. A small operator who answers their own phone said she was losing 40% of dinner-rush orders to busy signals before she added a text line.
    2. Aggregator fees are eating their margin. Owners on DoorDash and Uber Eats described commissions in the range of one-fifth to one-third of the ticket. SMS ordering is a way to claw a chunk of that back — not because the channel is free, but because the cost is fixed instead of per-order.
    3. Their customers were already texting them. Several owners said customers had been texting their personal cell phones for years. The chatbot just made it official.
    OJThe OrderJam take
    This category is real, but the implementations are uneven — we built OrderJam because most existing options are bolt-on chatbots that don’t actually integrate with the kitchen ticket printer. If you’re a restaurant evaluating SMS ordering, the question to ask isn’t whether it can take an order. It’s whether the order ends up where it needs to be.
    See how OrderJam handles this→

    The thing I keep thinking about

    After three weeks I caught myself avoiding restaurants that didn’t have a text line. That’s not a thoughtful editorial position — it’s a habit. And once a habit forms in a category like food ordering, the businesses that don’t accommodate it tend to lose customers gradually and then suddenly.

    The big chains will roll their own. The aggregators will try to extend into this category and probably succeed in the metro markets. But for the 700,000 independent restaurants in the U.S., the interesting question is which third-party platform they pick — because the labor of building it themselves is the same labor of building their own delivery network in 2014, and we know how that ended.

    I tipped the wing joint twenty percent. The bot thanked me. I think we’re going to be friends.

    #sms-ordering#ai#consumer-behavior#qsr

    Keep reading

    Why Restaurant Apps Are Quietly Dying
    Technology·6 min

    Why Restaurant Apps Are Quietly Dying

    A decade after every chain in America launched its own ordering app, app installs are flat, daily-active rates are tiny, and the channel that’s eating the share isn’t another app. It’s your messages thread.

    OrderJam Editorial · Apr 8, 2026
    AI in the Kitchen: What Works, What Doesn’t, and Where We’re Still Pretending
    Opinion·8 min

    AI in the Kitchen: What Works, What Doesn’t, and Where We’re Still Pretending

    A category-by-category audit of what AI is genuinely useful for in a restaurant in 2026 — and the four use cases the industry should stop talking about until the technology catches up.

    OrderJam Editorial · Mar 30, 2026

    © 2025 OrderJam. All rights reserved.

    Terms of Use