Give physical storesthe decision speed ofe-commerce.

Give physicalstores thedecision speed ofe-commerce.

Book a pilot conversation↓ Scroll to see how

01 · Now

You walk the floor.You hope you catch it.

A guest waits. A queue forms. A section goes cold. Today you find out when a manager happens to look up — or when the report lands tomorrow.

02 · Next

The brain catches it.The right call goes out.

Spark sees the floor across every camera, POS, and sensor at once. The instant something breaks, the right person gets the call — with full context.

03 · Future

Same brain.New hands on the floor.

Tomorrow, the same decision routes to a service robot. The brain doesn't change. Only what it dispatches to.

Spark · Spatial decision layer

01 · Now

02 · Next

03 · Future

SCROLL00%

TABLE A3 · 4:18 dwell · NO CONTACTENTRANCE QUEUE +5 IN 90SREFUND $48.20 · ROUTING TO GMCOVERAGE GAP · SECTION CREGULAR RECOGNIZED · MR. WEI · VISIT 11SHIFT 2/3 · 19:42 · DOWNTOWN LASERVER M ROUTED · ETA 22SB4 WATER REFILL · DISPATCHEDTABLE A3 · 4:18 dwell · NO CONTACTENTRANCE QUEUE +5 IN 90SREFUND $48.20 · ROUTING TO GMCOVERAGE GAP · SECTION CREGULAR RECOGNIZED · MR. WEI · VISIT 11SHIFT 2/3 · 19:42 · DOWNTOWN LASERVER M ROUTED · ETA 22SB4 WATER REFILL · DISPATCHED

01 · The gap

Same moment.
Two different worlds.

Every customer moment that an online store handles in seconds, a restaurant floor still handles after the shift — or never at all.

▣ Online store

When a customer…

▣ Restaurant floor · today

T + 5sExit-intent fires. Discount or chat pops.

Hesitates / waits too long

T + 4mStill sat at A3. Server is in section B.

T + 1mRe-targeted within the hour. Reason logged.

Leaves unhappy

T + 12mEmpty seat. No one tracks why.

T + 0sAuto-scale. Another checkout lane opens.

Pile-up at the door

T + 0s7 waiting. Staff still in cold sections.

T + 10sOne-tap, policy-checked, refunded.

Asks for a refund

T + 8mServer hunts for the GM mid-shift.

T + 0s"Hello again — your usual?"

Walks in as a regular

—New server. Doesn't know them.

Every physical store has the data.
None of them have the brain that turns it into action.

02 · How it works

One decision layer. Three phases.

Spark already runs in 32 stores doing awareness. Action is rolling now. Automation is next. The brain doesn't change — only what it dispatches to.

Catch problems during the shift, not after it.

Spark watches the floor across cameras, POS, and sensors. The instant something breaks — a table waiting too long, a queue spike, a section without coverage — it surfaces on your phone with full context.

Median time-to-surface

Live in

32 stores

FLOOR · SECTIONS

ENTRANCE

QUEUE · 2 waiting

COVERS

42+3

AVG WAIT

6.4m

TURNS / TBL

2.1

ALERTS

EVENT FEED

1 events · last 90s

19:42:08Section B2 — covers seated, 2 tops.

03 · On the floor

What e-commerce got right.
Now on your floor.

Online stores didn't win because they had better waiters. They won because they could test, measure, onboard, and stay consistent. Six of those same capabilities — running on your floor.

01A/B TESTINGTest the floor like a webpage↗ Try a new seating order, staffing model, or table-flip rule. Spark measures the lift in covers and turn time.lift per shift

02ANALYTICSFunnel the shift↗ See exactly where guests drop off — door, queue, seat, order, return. Like a checkout funnel, for your dining room.per moment

03ONBOARDINGUp to speed in one shift↗ New servers get live guidance — where to go, who's a regular, what's 86'd, what to upsell. Onboarding in real time.1 shift

04CONSISTENCYSame standard, every location↗ Service quality stops depending on which manager is on. Same brain runs every store the same way.32 stores · same

05PERSONALIZATIONRecognize every regular↗ Their last order, allergies, and seat preference follow them across the chain. Like Amazon — on the floor.cross-store

06ATTRIBUTIONAttribute every cover↗ End-of-shift, you know exactly what drove the night — what worked, what cost you, what to change tomorrow.< 30s · auto

04 · Live now

Running in 32 stores. Paid for. Used every shift.

32stores
live

Spark is in 32 locations of a regional hotpot chain in China — paying customer since April 2026. Managers use it during the shift, not after.

CustomerRegional hotpot chain · 32 locations · China

Live sinceApril 2026 · paid from day one

Pipeline2 additional chains negotiating pilots · US launch Q4 2026

“Before Spark, we knew which shifts went badly. Now we know which tables went badly — while there’s still time to fix them.”

Operations leadRegional hotpot chain · China (paraphrased)

05 · The longer arc

The worker changes.
The brain stays the same.

Restaurants are the first vertical because they're the loudest. The same gap between physical and digital exists in retail, fitness, hospitality, healthcare. Spark is the spatial decision layer underneath all of them. Today it dispatches your staff. Tomorrow it dispatches the robots that work alongside them.

VERTICAL 01

Restaurants

LIVE · 32 STORES

VERTICAL 02

Specialty Retail

VERTICAL 03

Fitness

2027

VERTICAL 04

Hospitality

2027

VERTICAL 05

Healthcare

LATER