Live MaxWay · running in 22 locations

Your cameras already see.Now they understand.

A guest walks in, the queue grows, a table frees up — your cameras already see it. SERVO turns that into a signal on your manager’s phone: on your own server, with no new hardware.

One location · three days · no cost

On-premise · on your own server Existing cameras · no new hardware No images stored · data never leaves the building
0
MaxWay locations
0
AI modules
24/7
control · without a human
0
contract to live

LiveReal recording · MaxWay Roison · entrance cameraNot a mockup: the running system’s own screen

Your shift today
The problem

248,000 UZS was recorded.
Everything else was not.

This check is the only record your business kept of that minute. Who the guest was, how long they waited, whether they ever come back — nowhere. An owner’s eight questions are born every day out of exactly these blank lines.

Location · Floor A02.07.202621:06
CHECK # 1042
Table07
Guests3
Lavash×296,000
Kebab×1110,000
Tea×342,000
TOTAL248,000UZS
Paid · cash
What it did not5 lines the check never carried
Who was the guest?Not recorded
How long did they wait?Not recorded
When did a server come?Not recorded
Which visit was this?Not recorded
Will they come back?Not recorded

The only record your system kept

8

questions, and not one of them has a number for an answer today.

“If you cannot see it, it does not exist.”

Eight questions a restaurant owner asks

Your answer today

Tap a question — we show why there is no answer

The proof

An ordinary working day. The footage speaks for itself.

Recorded at MaxWay locations. The date, time and camera number in the frame are unaltered — pause it, zoom in, check.

LiveCamera 02 · 17 Jun 2026 · 12:39Unedited footage

Every guest who walks in is recognised and their tier updates instantly.

Four recordings — pick any
Recognising one guest0.31s · total
Face vectorFound in the baseProfile opened

By the till, the guest’s tier is already on screen.

41ms
One frame analysed

11 modules — one frame, one server.

9
Active tracks at once

Nobody drops out, even at peak.

22%
Guests seen in two locations

More than one guest in five is already a regular at another of your locations.

MEASUREDMaxWay · 22 locationsEverything else is a model
Your restaurant

Where is your restaurant losing money?

Eight questions, each answered by the cameras you already have. Pick one: one restaurant, one day, and every number shows its working.

Guests

Who came through the door, and who comes at which hour?

Chilonzor · todayDemo data
Came through the door488counter 531 · staff −43
Returning guests18538% of guests today
Gold guests298 or more visits
Party size at dinner3.2people on average · lunch 2.2
Tier · today's guests
  • New30362%1 visits
  • Bronze10722%2–4 visits
  • Silver4910%5–7 visits
  • Gold296%8+ visits
Age × hour · who comes when
Who they come with · by part of the dayalonepairfamilygroup
Lunch46%30%2.2
Afternoon34%36%20%2.4
Dinner16%30%38%16%3.2

At lunch the main guests are aged 25–34, alone or in pairs. In the evening the 18–24 group grows fastest, and 38% of tables are families. 51% of guests are women.

Decision

Run a quick solo menu at lunch and family tables and set menus at dinner. The host sees Gold guests the moment they walk in.

01Guest Recognition02Customer Segments04Daily Traffic

What this costs in money is below, worked out from your own daily guests and average check.

Work out the cost
The step after the console

Upload the report — get the answer

The console exports last night as spreadsheets. Type your question — SERVO's analyst reads those files and answers: one figure, the evidence for it, and one next step.

SERVODEMO

Why did tables sit empty last night?

Tables were not cleared as guests left: for 86 minutes a table stood unsold, and that is exactly when 5 parties walked in.

  • 86minutes a table stood idle past the 5-minute line
  • 5parties walked in during exactly those minutes
  • 4guests found no seat that evening
What caused it, and what was ruled out

Tables stood uncleared for 141 minutes in all last night; 86 of those are past the 5-minute service line. An average seating runs 64 minutes, so that is ~4 guests — about 120 a month at the same rate.

Mechanism. At 20:18 the floor had 3 uncleared tables and 3 free ones. The longest was table 14: 28 minutes. The signal already exists — the console writes “table 09 is free”; the question is whether the message reaches anyone.

  • The data rules out
  • “The room was full” — it was not: at that minute 3 tables stood empty and 3 uncleared.
  • “The waiters could not keep up” — the evening ran 68 seatings, a 3.4× turn; what slowed were 11 clear-downs, the ones where the message was late.

Conclusion. This is not a headcount question but a signal question: the minutes between “table free” and “table clear” are not sold.

How it is measured. 14 days: the median clear-down time — the count of clear-downs over 5 minutes should fall towards zero; the control metric is the number of parties kept waiting.

Next step

The moment a table empties, send the signal to the floor manager's phone and close the clear-down within 5 minutes — the “table 09 is free” message is already in the console.

  • WhoFloor manager · clearing staff
  • WhenFrom tomorrow, 18:00–23:00
  • Expected result~4 guests an evening, ~120 a month
Files read:stollar.xlsxtashriflar.xlsx

The system

Eleven modules. One system.

Each module answers one question on the floor. All of them run on one data layer — which is why the system reports not that there was a queue, but why there was one.

  • 05Table Occupancy03Customer Traffic08Speed of Service

    "At 20:10 three people were queuing — because tables 7, 9 and 12 sat uncleared for nine minutes."

  • 01Guest Recognition02Customer Segments10AI Forecast

    “Your Gold guest has fallen out of their fourteen-day rhythm — third week without a visit.”

  • 11Safety and incidents09Cleanliness & Tables07Staff KPI

    "The fryer was unattended for four minutes — two staff went on break at the same time."

Each of the three sentences needs all three of its modules; with two of them it is never produced at all. A queue counter reports that there is a queue. Only a system whose tables, footfall and service clock sit in one data layer reports why.

The module catalogue

11modules · open any cardModel
Entrance · recognition01Guest RecognitionFirst visit or tenth0,31 sto recognise one guest
The base · by tier02Customer SegmentsRegulars and newcomers+23%a regular’s check
Visits · by hour03Customer TrafficThe busiest hours−20%queue walkaways
Visits · by age04Daily TrafficWho your floor serves5 109guests a day · MaxWay
Floor · 12 tables05Table OccupancyIdle table minutes19 → 4minutes — table clearing time
Zone × hour06Table HeatmapWhat each metre earns11%dead-zone occupancy
Time with guests07Staff KPIWaiters at the table−9%in wasted payroll
To first contact08Speed of ServiceTime to first service30–60 sindustry norm
Shift check09Cleanliness & TablesKitchen hygiene checks0:12from breach to reminder
Flow · measured and forecast10AI ForecastTomorrow's prep list−30%food waste
Zones under watch11Safety and incidentsFire and conflict alerts0:10from incident to alert
The money

Three numbers. One answer.

Guests a day, the average check, the number of branches — nothing else. Every line shows its arithmetic.

60–1,600
25,000–400,000
1–30
Your daily turnover48 M UZSguests × check × locations
Where it goes todayPer daySERVO recovers

0.50% of daily guests × average check

Over the queue threshold an alert reaches the manager and a table is prepared

03 · Customer Traffic
Total, every day2.6% of your turnover — where no check is written1.3 M UZS399 K UZS
Recovered per month
12M UZS / mo

You pay per month 3.1M UZS · Net 8.8M UZS · 3.8×

Breaks evenbase 1.0 monthscautious 2.6 months+95 M UZS0 mo3 mo6 mo9 mo12 mo
2.6%of turnover — slips away0.2%of turnover — the platform
Recovered over the year
+136.6
Subscription · 12 mo
−37.8
Setup · one-time
−3.8
Net, first year
+95.0 M UZS
We will measure one location, free

If the daily report is not running within 30 days, you get a full refund.

What you are really choosing between5.0one more manager4.0a loyalty app plus integration1.5mystery shopping · monthly3.1SERVO · all eleven modulesEach of these closes a fraction of the eight lines above. Tashkent market estimate.
The model’s assumptions — all of them
  • Loss shares are SERVO’s industry estimate; during the pilot they are replaced with your location’s measured figures.
  • The “SERVO recovers” column is the share of each line that actually closes (20–70%). An alert does not stop every walk-out, and a dead zone needs the floor plan changed.
  • Platform cost: $250 per location per month for the standard package, at 12,600 UZS to the dollar. Real pricing is quoted per requirement.
  • Month one runs at 55% and month two at 85% while thresholds settle to the kitchen. The break-even figure is computed with that.
  • One-time onboarding: camera survey, edge server, POS integration and team training — $300 per location.
  • Cautious case: every line recovers half as much. That is the dashed line on the chart.
What does not changeNo camera is replacedNobody is laid offNo process is broken
Pilot

Get the numbers you just set, in writing.

8.8M UZSreturned per month
3.8×per 1 UZS of fee

1 location · 420 guests/day · ø115 k UZS check

The figures come from the table above: each of the eight leaks is priced separately, and the platform fee is subtracted. See the arithmetic

We usually reply within one business day. No spam.

Where the data stays

The face is not stored. Only a number remains.

The system does not need a picture to recognise a guest. A mathematical vector is taken from the frame and the frame itself is destroyed at once — the matching runs on the vector.

YOUR RESTAURANTCamerasframeframe destroyedEdge serveroffline, 24/7vector onlyYour dashboardNo face crosses this lineA cloud abroadFace · video
1The camera takes a frame2It becomes a vectorThe frame is destroyed0 images kept
The life of one frameA cloud vendorSERVO
Where the AI runsThe vendor's cloud, abroadYour server, inside the restaurant
Where the video goesUploaded or streamed outIt never leaves the building
How a guest is recordedA personal profile in the cloudNo name — an ID and a loyalty tier
The Republic of Uzbekistan’s Personal Data lawRetention is set in the contract, agreed with your lawyers before go-liveAnonymous mode: recognition off, every other module still runsAudit log: who opened which data, and when
Implementation Plan

Live in three days. Two hours of yours.

Every day in two lines: what that day takes from you, and what you hold by the end of it. On none of the three days does the floor close.

  • 3 daysfrom contract to live
  • 2 hoursof your time, in total
  • 0new cameras — yours are used
  1. Day 1

    Audit and install

    From you
    Access to the cameras
    You have
    A camera-coverage map and the edge server in place
  2. Day 2

    Tuning and integration

    From you
    Nothing
    You have
    The system in sync with POS / 1C, SOP thresholds set to your kitchen
  3. Day 3

    Training and go-live

    From you
    Two hours
    You have
    That same evening — the first daily report. From then on, 24/7 monitoring.
After go-live
  • 24/7 monitoring — we watch the system
  • A daily report every evening
  • No report within 30 days — full refund

Which location do we start with?

The busiest one — that is where the numbers show up fastest. In a month you put the second one beside it.

  • First call — 15 minutes
  • From you: a camera list and two numbers
  • No sales pressure
  • One training session for the team, no extra work for you