Palate is the cross-platform dining identity that every AI assistant, every restaurant, and every future interface will need to call. Built from the one angle with the full picture — the diner's.
Consumer software is entering a period of interface substitution. Apps are being joined, and in many cases replaced, by AI assistants that handle discovery, decisions, and bookings on the user's behalf. OpenTable has partnered with ChatGPT. Resy is live in Claude. DoorDash and Uber Eats are rolling into OpenAI's assistant. Every major dining platform is becoming an AI-addressable surface.
This is one half of the shift. The other half is that restaurant tech itself is consolidating. OpenTable just updated its terms to require restaurants to make it their "system of record." Resy and Tock are merging under American Express. DoorDash acquired SevenRooms and is now a reservation platform. The walls between platforms are getting taller, not shorter.
The category is moving toward a state where interfaces are fungible and platforms hold their data tighter than ever. In that environment, the value shifts down the stack — to what every interface has to call in order to be useful. The context. The identity. The memory. The decision-making fabric.
That is the layer Palate is built to own.
Every dining transaction a person creates generates a record. A reservation confirmation from OpenTable. A delivery receipt from DoorDash. A booking email from Resy. A POS check from Toast. Every single one of these records lands in the diner's own inbox. No platform sees across them. The diner does.
Palate's mechanism is to enlist the diner as the connector. The data the category has been unable to stitch together across platforms has been sitting, all along, in a place no one was looking: the diner's email. Palate reads the records, deduplicates them, resolves them into canonical restaurants, and builds the dining identity only the diner has ever been positioned to own.
A sophisticated reader will ask whether a well-funded integrator could build a mega-CRM that connects to OpenTable, DoorDash, Toast, and every other platform directly. In theory, yes. In practice, it still falls short — because even a mega-CRM would be assembling fragments from platforms that don't actually know who their customers are. A delivery platform knows "Jon M. ordered on April 11." It does not know that Jon M. is a specific identity with a full cross-platform history. Only the diner can confirm that. Any approach that does not enlist the diner will keep assembling fragments and calling the result a profile.
This is not a new pattern. Shopify's Shop app does the equivalent thing for e-commerce: it aggregates purchase history across merchants by going to the shopper, not the merchants. No retailer would ever hand Shop its customer data — but a shopper who installs Shop and connects their email gives Shop access to everything they have bought, everywhere, because those confirmation emails all land in their own inbox. Shop turned that access into a consumer product people actually use. Dining hasn't had this move yet. Palate is the answer.
Palate is a layered platform. Each layer has a specific job. The value accumulates in the foundation and extends outward through the connectors. Losing a single surface doesn't lose Palate — new surfaces can be added without rebuilding.
Every new diner enriches the data layer. Every new platform adds a parser. Every new connector extends the reach of the intelligence engine. Every new surface gives the platform a new place to show up. The architecture is designed so that time works in its favor.
Palate creates a view of a diner's taste — their palate — in the same way Spotify created a view of a listener's music taste. What they eat, where, when, how their taste has shifted over the years. All of it, stitched and computed into a living profile that belongs to the diner.
Within minutes of connecting, a diner is looking at hundreds of dining occasions they had forgotten. Cities they passed through. Restaurants they loved. Patterns they didn't know they had. It's the first time most people see their own dining life as a complete picture.
Thursday 6:15pm. Palate knows it's usually pizza night. Knows the three new spots in town that would earn the slot. Every recommendation explains itself. Every suggestion is drawn from the full picture, not a slice.
Palate doesn't wait to be asked. A Friday morning nudge about Saturday reservations. A travel-city prompt when the diner's location changes. Patterns, once recognized, surface at the right moment without effort.
Connect Palate to Claude or any MCP-compatible assistant. The assistant stops asking what the diner likes and starts answering. Where to eat. Who it's for. What fits. With live availability surfaced inline and one-click completion on the native booking platform.
For the first time, restaurants can see the diners at their tables as complete people — not first names and party sizes. Palate gives operators a cross-platform view of every guest, every diner they haven't yet met, and the intelligence to act on both.
A restaurant that sees a diner twice a month may assume loyalty. Palate shows whether the restaurant owns 30% of that diner's dining life or 5% of it. A regular who gives you 30% is a different asset than a regular who gives you 5%. For the first time, that's a number.
Every third-party delivery order is a customer a restaurant paid 15–30% commission to serve and then lost. Through the diner's own permission, Palate turns the anonymous “Jon M. ordered on April 11” into a real person the restaurant can reach directly. The single most painful structural problem in the restaurant business, solved.
"Italian eaters within 20 miles who dine out 5+ times a month and haven't been to us." Acquisition by actual behavior, not demographics. No other platform can deliver this — because no other platform has the cross-platform view.
When a loyal guest's overall dining is growing while their visits to your restaurant stay flat, they're slipping away slowly. Palate surfaces that signal early, often with context about where their attention has moved.
Third-party marketplaces deliberately withhold the customer. Palate routes around that with the diner's own permission. Here's what a restaurant sees the moment a DoorDash order lands — before and after.
The initial on the ticket becomes the same diner, fully named, with a complete dining identity the restaurant can actually build a relationship with. The single most painful structural problem in the restaurant business — solved at the source, through the diner.
Palate is live in Claude today at thepalate.app/mcp. A diner who connects Palate to any MCP-compatible assistant gives that assistant their full dining identity — cuisines, patterns, geographic eras, the works. The assistant stops asking what they like and starts answering. Seven tools today; more coming. The arc is from recommending to doing: live reservation slots surfaced inline, one-click completion on the native platform, with full in-conversation booking coming next.
The tool set is a starting sample. The full platform surface will expand as more capabilities come online.
Every number here comes from a single diner's live history, ingested through the production platform today. The evidence is the platform running. No staging data, no projections.
The dominant platform in this diner's history holds just 29% of it. Any approach built from inside one platform misses the majority of the picture.
The platform is live at thepalate.app. The AI surface is live at thepalate.app/mcp. The architecture is in place. The data is flowing. The profile computes. The assistants can call it.
The work ahead is expansion — more diners, more platforms, more surfaces, deeper intelligence. Not fundamental construction.
The concept is born.