DoorDash Launches Zesty — and Quietly Reveals What It Bought SevenRooms For

I was halfway through a flat white at a coffee bar on Valencia Street this morning when a friend who runs operations for a small SF group slid his phone across the table and said, “look at this — it’s actually good.” The app on the screen was Zesty, the AI restaurant-discovery product DoorDash had just put into beta a few hours earlier. He was using the chatbot to argue with it about whether to try a new Cal-Italian place in Hayes Valley or the omakase counter he’d been meaning to revisit. The bot was holding its own.

That was the moment the SevenRooms acquisition started to make sense to me in a way the press release never did.

The contrarian read on today’s news is the one DoorDash isn’t saying out loud: Zesty is not really a discovery app. It is the front end of a reservations-and-CRM platform DoorDash spent $1.2 billion to buy in June, and today is the first day the public can see the shape of what they’re building. The discovery layer is the wedge. The booking and the guest-data sit underneath.

The wedge nobody is calling a wedge

Here’s what DoorDash actually shipped today, per the TechCrunch piece that broke the launch. Zesty is in beta in the San Francisco Bay Area and New York City only — not national, not even a top-ten-metros roll-out. The product is a chatbot you talk to about where to eat. It aggregates signals from DoorDash’s own data, Google Maps, TikTok, Reddit, and Eater. It surfaces a “Top Zesties” leaderboard for places that are trending across those feeds. Co-founder Andy Fang announced it himself, which tells you the founders are still close enough to the product to put their name on it.

What’s interesting is the list of things Zesty doesn’t do today. It doesn’t book the table. It doesn’t store your dietary preferences across visits. It doesn’t tell the restaurant you’re coming. It doesn’t follow up after the meal. Those are SevenRooms verbs, every one of them — and SevenRooms is sitting one floor down in the same building, integration-wise.

So the question I’d ask any DoorDash PM honest enough to answer it: what is the natural product evolution here? You build a discovery surface that gets users into the habit of talking to a bot about restaurants. You learn what they ask for, what they reject, which neighborhoods they roam, which cuisines they cycle through. You earn the right to add a booking button. And the booking button, when it appears, is powered by the platform you already own. Discovery on the front. Reservations and CRM on the back. The wedge is the chatbot.

Why SF and NYC, and why now

The geographic restraint is the second tell. DoorDash holds roughly 67% U.S. delivery share per Deliverect’s category tracking, which means they could have launched this product in fifty markets on day one and DoorDash users would have shown up. They chose two. Both are dense, both are restaurant-saturated, both have the kind of independent-operator base where SevenRooms already has a defensible install — and crucially, both are markets where OpenTable and Resy are entrenched enough that a head-on assault would be expensive.

A two-market beta lets DoorDash learn the bot in the cities that matter most for their reservations ambitions without setting off the OpenTable alarm bells. By the time Zesty is in twenty markets, the booking layer will be live, the SevenRooms integration will be visible to operators, and the conversation in the category will have shifted from “another discovery app” to “DoorDash just turned reservations into a distribution game.” That is a forthcoming Bottom Line piece, and the math on the $1.2 billion will look very different once the wedge is fully extended.

The reservations incumbents are not unaware of this. OpenTable has been quietly seeding AI into its own product surface — a forthcoming May Pass piece will get into the technical details — and Resy has its own moves planned through the Amex relationship that an upcoming spring piece will unpack. But the question I keep coming back to is whether either of them can match the consumer-side gravitational pull DoorDash already has. OpenTable’s discovery surface is dinner-tonight transactional. Resy’s is curatorial. Zesty’s is conversational and AI-native, and the underlying customer base it can pull from is two orders of magnitude larger than either competitor’s reservation book.

Mark interpretation

What I think is actually happening today is the public unveiling of a strategy DoorDash has been telegraphing for eighteen months but never quite saying. The June acquisition announcement framed SevenRooms as a play to “battle OpenTable,” and the framing was right but the mechanism was wrong. You don’t beat OpenTable by being a better OpenTable. You beat OpenTable by changing the surface where restaurant decisions get made — moving it upstream from the booking page into a conversation — and then owning the booking layer when the user gets there. Zesty is that upstream move.

There’s a commerce angle here too that I’m watching but not yet ready to fully call. DoorDash has been building out the broader merchant-commerce stack for two years, and the May DoorDash commerce piece will revisit how Zesty fits into the larger flywheel — delivery, reservations, gift cards, loyalty, ads, the whole thing. The short version: a discovery surface that knows what you want for dinner is also a surface that knows when to serve you an ad, when to push a delivery option if the restaurant is booked, and when to upsell a loyalty enrollment.

Two things to watch from here. First, when does the booking button appear inside Zesty — and is it SevenRooms-powered or a more neutral integration meant to keep operators on multiple reservations stacks happy? Second, how do operators currently on Resy and OpenTable read the Zesty announcement? My guess is the smart ones are already asking their reservations rep what their distribution looks like inside DoorDash’s discovery surface, and the answer is going to be uncomfortable.

The launch is in beta. The market is two cities. The product is a chatbot. None of that is the story. The story is what gets bolted on next.

— Samuel hosts the Service podcast for TableTransfers. Tips: [email protected].