PeekaPlate turns a restaurant’s own photos into 3D dishes guests explore while they wait, then shows owners which dishes were considered but not ordered, and why.
Every table leaves a trail of decisions. The POS only keeps the last one. PeekaPlate captures the rest and turns it into menu changes owners can test.
Everything a guest considered before ordering disappears, along with the reasons they changed their mind.
Price, portion, photo or placement? Menu changes run on gut feel and sales totals.
Guests can’t picture unfamiliar dishes, and the 10–20 minutes on the waitlist go to scrolling something else.
A QR at the host stand opens the menu in the browser. No app download, no account.
waitlist_join · party 2Rotate, zoom and compare real-scale dishes in 2.5D, 3D and AR, built from the restaurant’s own photos.
dish_view · harvest_bowl · 3dViews, dwell, compares and saves become anonymous, structured events attached to each dish.
dwell · harvest_bowl · 14,200msSignals feed a weekly owner insight, one-click menu experiments and, with opt-in, follow-ups.
experiment_start · harvest_bowlA QR at the host stand opens the menu in the browser. No app download.
waitlist_join · party 2
Rotate, zoom and compare real-scale dishes in 2.5D, 3D and AR, built from the restaurant’s own photos.
dish_view · harvest_bowl · 3d
Every view, dwell, compare and save becomes an anonymous, structured event attached to the dish.
dwell · harvest_bowl · 14,200ms
Signals feed owner insights, one-click menu experiments and, with opt-in, follow-ups.
experiment_start · harvest_bowl
High consideration, long dwell, low purchase.
Guests open the dish, zoom into the bowl, then go back to the list.
Likely doubt about portion size or value for the price.
Show a clearer portion view, or pair it in a bundle.
A/B test: portion view vs. bundle offer. Measure menu-item conversion.
Each dish comes apart layer by layer, built from the restaurant’s own photos. Every turn, zoom and pause is a signal of what the guest wants to know before they order.
Which dishes get considered and which get ordered, item by item.
Test bundles, add-ons and upgrades against real intent.
Guests choose sooner with clearer dishes and portions.
Segments built from what guests explored, not only what they bought.
Which promotions move consideration, and which move orders.
These are the capabilities we will measure in pilots. We don’t quote performance numbers until pilot data exists.
Waiting guests are on their phones anyway. A QR preview meets them there, with nothing to install.
Image-to-3D tools turn a restaurant’s own photos into 3D dishes in minutes, not a studio shoot.
86% of operators are comfortable with AI tools; 26% already use AI for menu optimization.
Toast AI survey, 2025 · n=712Snap plans consumer Specs for 2026, Google and Samsung Android XR glasses are expected in fall 2026, and Meta keeps expanding its glasses. The same menu data and intent graph work on phones and glasses.
Reuters, TechRadar, Meta Connect| 3D / AR menus | QR ordering & POS | Waitlist & CRM | PeekaPlate | |
|---|---|---|---|---|
| What it sees | The visual effect | The final order | The visit and past purchases | What guests considered before ordering |
| Explains why a dish underperforms | – | – | – | Yes, per dish |
| Runs the fix | – | Sometimes, from order data | Generic promotions | Menu experiments and intent-based follow-ups |
| Works with your stack | Standalone | Is the stack | Is the stack | Connects to POS, waitlist and CRM |
We start with the 10–20 dishes that drive most orders at visual-first restaurants: poke, ramen, dessert, cocktails. Each table adds to a graph of what guests want before they buy. Phones today, glasses tomorrow.
The menu is the beginning. The intent graph is the product.
Model the top dishes, place the QR card, connect the POS order report.
Anonymous browsing per dish, matched daily to what was actually ordered.
Flag high-attention, low-order dishes; run one A/B test on the flagged dish.
A readout for the owner, and a go/no-go on the model.
Biggest risk: matching sessions to orders. Pilot fix: a manual daily match with the POS report.
You run a visual-first restaurant where looks sell dishes
Guests regularly wait 10+ minutes for a table
You want to know why some dishes don’t sell, and test a fix
You want to replace your printed menu at the table
You need full-menu 3D on day one
You don’t use a POS that can export order reports
No. A QR code opens PeekaPlate in the phone’s browser. AR uses the phone’s built-in viewer (iOS Quick Look, Android Scene Viewer).
No. We start with the restaurant’s existing photos of its top 10–20 dishes and generate 2.5D and 3D views from them. Real scans can come later.
Browsing is anonymous and kept separate for each restaurant. A phone number is only linked if a guest opts in to follow-ups, every text includes STOP to unsubscribe, and we never message about dietary or allergy information.
No. PeekaPlate is an optional preview while guests wait. It connects to the POS, waitlist and CRM tools a restaurant already uses.
[Pilot pricing to confirm.] Early pilots are designed to be low-effort for the restaurant: a QR card, your photos and a daily POS report.
We’re onboarding our first pilot restaurants in San Francisco and talking with early investors and operators.
hello@peekaplate.io →