See every step of service.
At every table. As it happens.

LAB Optic is an optical service-execution layer for restaurants: an objective, timestamped record of what the house did during the shift, tied to the revenue it protected or lost.

No POS, reservation system or guest survey can produce this record. LAB Optic does, during the shift, by section and by position.

The problem

Guest experience fails on execution, not on menu or design.

A late greet. A manager who never visits. A plate set down wrong. A table left unbussed. Each is invisible in the P&L, each is decided in minutes, and each drives whether the guest returns.

Today management learns about these misses days later through reviews, if at all. By then the table has turned, the shift has closed and the guest has already decided.

LAB Optic changes the timing. Ceiling cameras running on-device vision models detect each step of service as it happens and record whether it met the house standard. A miss becomes a discrete, attributable fact: which table, which section, which position, how late.

The result is a service record management can act on during the shift, and an empirical link between service execution and repeat visits that no operator has had before.

What it sees

The steps of service, as events.

LAB Optic detects events and dwell, not identity. Staff are recognized by uniform and role zone. Every event carries a table, a section, a role and a timestamp.

01

Seat

Party seated; the service clock starts.

02

Greet

First staff approach to the table, timed from seating.

03

Order

Server dwell at the table in order-taking posture.

04

Beverage delivered

Drinkware placed at each cover.

05

Food delivered

Plates placed at each cover, timed from order.

06

Plate placement

Position checked against the house placement standard.

07

Manager touch

Manager presence at the table during the entrée window.

08

Check-back

Server return within the window after delivery.

09

Clear

Plates removed after the last guest finishes.

10

Check drop & close

Presenter placed, retrieved, party departs.

11

Bus

Table cleared of all service items.

12

Reset

Table re-laid and ready for the next party.

The event set is configured per concept. Fine dining adds crumb-down and wine presentation; fast casual drops manager touch and check-back. The detection method is the same in every case.

What it measures

Every step against your standard.

Each event becomes an interval measured against the operator’s own service standard. Two composites roll it up: the Service Execution Score, the single number a GM reads at pre-shift, and the Table Miss Count, which flags a table live so a manager can intervene before the check drops.

Metric Definition Standard

Time to greet

Seat to first staff approach

Set by operator

Time to first beverage

Order to beverage delivery

Set by operator

Ticket time

Order to food delivery, per course

Set by kitchen

Manager touch rate

Tables visited by a manager in the entrée window

Set by operator

Check-back compliance

Server return within window after delivery

Set by operator

Placement compliance

Plates set per placement standard

Set by operator

Clear and check latency

Last guest finished to cleared; cleared to check drop

Set by operator

Bus and reset time

Departure to table ready

Set by operator

Dead-table time

Occupied minutes with no staff contact beyond standard

Reported

Inside LAB Platform

The execution layer under the LAB Panel.

CPLH tells you whether the floor was staffed to plan. LAB Optic tells you whether the staff on the floor executed. A section running CPLH at 1.3 with a 40% greet miss is a training problem, not a scheduling problem, and the Panel alone cannot tell the two apart.

LAB Convergence

Service events land as a connected data source beside POS, labor and reservations, in the same ontology. One Truth.

LAB Signal

New system-initiated triggers: live table flags, section miss clusters by day-part, position-level patterns, before the week closes.

LAB Converse

"Why did section 3 turn slow Friday?" answered from the event record, not from recollection.

The revenue bridge. Tables with misses and tables without are compared on same-visit outcomes from the POS (tip rate, dessert and beverage attach, average check) and on return behavior from the reservation system (repeat rate at 30, 60 and 90 days). After one to two quarters an operator holds an empirical cost per miss type, calculated, not estimated, and it feeds LAB Signal as an EBITDA driver alongside RevPASH, CPLH, Prime Cost and FixPASH.

Architecture & privacy

No biometric identifier. By design, not by policy.

Frames are processed in memory on an edge unit inside the restaurant and discarded. Only the event record leaves the device.

Status

In development.

LAB Optic is an AnalysisLAB system under active development. Specifications on this page describe the design intent and are subject to change as the pilot program is defined.

Now

Concept & design

Event set, standards framework and edge architecture defined. Privacy position under counsel review.

Next

Pilot house

Single-section deployment with a LAB Platform operator. Standards set by the operator; revenue bridge run on their own POS and reservation data.

Then

Platform release

Service Execution Score in the LAB Panel; LAB Optic triggers in LAB Signal; multi-unit rollout.

Early access

Operate the first house.

Pilot participation is limited to LAB Platform operators. If your group runs a service standard and wants to prove what it is worth, start the conversation.