The dining data layer

Interfaces are becoming substitutable. Data layers are not.

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.

Dining identity
Maria C.
Bay Area · since 2011
1,110Occasions
343Restaurants
55+Cities
16Platforms
OOpenTable322
RResy184
TToast291
DDoorDash146
GGrubhub52
CCaviar41
SSquare33
CCredit63
01 — The Shift

Two forces are converging. The category hasn't adapted.

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.

Interfaces multiplying
ClaudeChatGPTGeminiMCP assistants
Each one plugging into a single dining platform's slice.
Platforms consolidating
Resy + Tockmerging under American Express
DoorDash + SevenRoomsdelivery becomes a reservation platform
OpenTablenew terms: "system of record"
Walls between platforms getting taller, not shorter.
02 — The Insight

Every incumbent is building a profile from the slice they control. Only one angle has the whole picture.

The diner is the only connector

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.

The mega-CRM objection

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.

The Shop parallel

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.

Today
With Palate
What each platform sees
Fragments. The top platform holds just 29% of the picture.
fragments
OGallina d'Oro bookingApr 11, 2026
DRubirosa takeoutApr 09
TTartine — check $48Apr 07
RKato SF reservationApr 05
The objection:a mega-CRM could stitch this. But it's still fragments of identities nobody actually knows. Only the diner can confirm who they are.
What the diner sees
One identity. 16 platforms. 15 years. 100% of the picture.
complete
M
Maria C.
Weeknight · Italian-leaning · Bay Area
1,110
Occ.
343
Rest.
55+
Cities
15
Years
Share of diner's dining life, by platform
Toast
29%
OpenTable
22%
DoorDash
12%
Resy
11%
Credit card
9%
12 others
17%
03 — The Platform

Five layers. Value compounds as they stack.

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.

Layer 01Data Ingestion
Email parsers for 16 platforms. Photo receipts. Direct integrations. Self-learning drift detection. Multi-channel by design.
Layer 02Data Layer
Canonical restaurant resolution. Occasion dedup across platforms. Fifteen years stitched into a single coherent timeline.
Layer 03Intelligence Engine
Cuisine affinity, occasion modes, geographic eras, taste evolution, share-of-wallet. Computed from evidence, not quizzes.
Layer 04Connectors
MCP server + programmatic API. Permissioned, scoped, revocable — the diner controls every client.
Layer 05Surfaces
Consumer web app, restaurant dashboard, AI assistants, partner integrations. Every interface the data layer can reach.

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.

04 — For Diners

You know your taste. Now your taste works for you.

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.

01

The onboarding reveal

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.

02

Recommendations grounded in real behavior

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.

03

Proactive intelligence

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.

04

An AI assistant that already knows

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.

9:41
Saturday
This Saturday, 7:30pm
Three spots that fit you.
Italian-leaning · $60–90 · Bay Area
94
Troubadour
New Italian · Bay Area
Matches your top 3 Italian spots
88
Bravas Bar de Tapas
Weekend regular · Bay Area
Last visit Jan 18
82
Valette
Cal-New American · Bay Area
Price band + cuisine fit
9:41
Profile
M
Maria C.
Italian-leaning · Bay Area · since 2011
1,110
Occasions
343
Restaurants
55+
Cities
Top cuisinesAll
Italian
92
Cafe
88
Pizza
76
Mexican
64
American
52
RecentSee all
APR 11
Gallina d'Oro
Dinner · 2
O
APR 09
Rubirosa
Delivery · $43
D
APR 07
Tartine
Walk-in · $48
T
Profile
Discover
Friends
Saved
05 — For Restaurants

You see a Friday regular. We see their other twenty-five meals.

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.

Guest IntelligenceAcquisitionRetentionReports
Last 90 days
MC
Known guests
1,284
+18% vs last Q
Share of wallet
21%
−3% vs market
At-risk this week
47
flagged early
Recovered from 3P
412
+91 this month
Share of wallet across the market
Your guests' dining, broken down by where it actually goes.
All cuisines
Us21%
Osteria 4218%
Rubirosa14%
Tartine11%
Rest of market36%
Guest roster
Hover to see their other dining life.
AllVIPAt riskGrowing
GuestShare of walletSpendStatus
M
Maria Chen
17 visits · last Fri
$2.1k
Regular
D
David K.
38 visits · Italian-leaning
$2.9k
VIP
S
Sam Patel
Weekly → monthly
$890
At risk
L
Lena Ruiz
New in last 60 days
$1.6k
Growing
D
David K.
38 visits · Italian-leaning
24%
share with us
15
other meals / 90d
Where else they eat
Osteria 427
Rubirosa5
Tartine (coffee)10
5 others0
Cuisine affinity
ItalianPizzaCafe
01

Share of wallet, finally visible

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.

02

Recovering the customers the marketplaces hide

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.

03

Finding the diners they want

"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.

04

Churn signals before they show

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.

What changes at the order

Same diner, same $24.50 pizza. The person behind it becomes visible.

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.

Today · via DoorDash
Guest
Maria C.
Contact
This order
1 Margherita · $24.50
Visits with us
Share of wallet
Cuisine affinity
Can you reach them?
no
With Palate · shared with permission
Guest
Maria Chen
Contact
maria.chen@palate-share.com
This order
1 Margherita · $24.50
Visits with us
4 visits · last Fri
Share of wallet
24% · VIP
Cuisine affinity
Italian · Pizza · Cafe
Can you reach them?
yes · with permission

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.

06 — The AI Layer
Live at thepalate.app/mcp

Every assistant can know who a diner is, through one connection.

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.

getProfile()Cuisine affinity, behavioral patterns, price bands
searchHistory()Query 15 years of dining by cuisine, platform, city
discoverRestaurants()New spots grounded in the diner's real profile
checkAvailability()Live Resy + SevenRooms slots, surfaced inline
getHistoryByCity()What the diner already knows and loves, per city
getPreferences()Stated dietary restrictions & overrides
updatePreferences()Set preferences without leaving the assistant

The tool set is a starting sample. The full platform surface will expand as more capabilities come online.

Claude · via MCP
Palate
Connected · 7 tools · as Maria
● Live
Where should we eat this Saturday? A friend is in town.
C
getProfile
weeknight vs weekend · cuisines · price band · areas
getHistoryByCity(city: "Bay Area")
42 visits · 18 restaurants · dominant: Italian, Cal-New American
Based on your weekend pattern — Italian-leaning, $60–90 range, in your area — three that fit. One you'd know, two you wouldn't.
94
Troubadour
New Italian · matches 3 of your top spots · $70–90
88
Bravas Bar de Tapas
Weekend regular · last visit Jan 18 · $55–75
82
Valette
Cal-New American · price fit · haven't been · $80–95
Troubadour sounds perfect. 7:30 for two.
C
checkAvailability(restaurant: "Troubadour", date: "Apr 4 2026", party: 2)
Resy · 7:30 unavailable · 4 alt slots
7:30 is booked. Live slots for Saturday:
6:45
7:00
8:15 ✦
8:45
8:15 fits your weekend dinner window. Here's the Resy link — takes 10 seconds to confirm.
Perfect — opening Resy now.
07 — Proof

Not a thesis. A running pipeline.

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.

29%

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.

Platform distribution · this diner's history
Toast29%
OpenTable22%
DoorDash12%
Resy11%
Credit card9%
12 others17%
1,110
Dining occasions
343
Restaurants
55+
Cities across 7 countries
16
Platforms represented
15
Years of history (2011 to today)
63%
Share held by the top three combined
Platforms represented
OpenTable · Resy · Toast · DoorDash · Grubhub · Caviar · Seamless · Square · Spoton · Slice · ChowNow · Reserve with Google · Postmates · credit card · receipt photo · order-direct
7 tools live
at thepalate.app/mcp

The dining data layer, running today.

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.