OBSRVR

We turn any camera into
leasing, retail and
security intelligence.

UAE · Seed Round · Confidential · September 2026
02

Everyone counts people. Counting was never the product.

Recognition is: it excludes staff, tracks journeys, qualifies attention, and — with opt-in — turns a mall into an ad platform with zero bots.

03

Richemont couldn't count
fifty customers a day.

Cartier, Van Cleef, Piaget: every boutique has a doorman, and the doorman stands under the ceiling sensor, triggering it all day. At fifty to a hundred customers a day, every miscount moves the numbers that pay the staff.

The legacy answer

Ceiling sensors and person detectors. They count everybody: staff, children, the doorman, twice.

Ours

Counting built on facial recognition. Staff and children excluded, individuals tracked, accuracy luxury retail can act on.

Why nobody copies it

Person detectors are a forked open-source model, and they inherently cannot track individuals. Recognition can.

04

The same cameras climb
four layers of value.

Each layer uses the one below it. Competitors are stuck at the first.

Count
Staff and children excluded: accuracy legacy sensors never reach
Track
The same faces store to store: journeys, dwell, brand synergies
Qualify
An impression counts only when the eyes are on the ad
Monetize
Opt-in identity via the mall's loyalty app: perfectly timed media, zero bots

Layer four is contracted as phase two with Majid Al Futtaim: customers opt in with their face through the loyalty app, and the operator sees the journey end to end, not just the purchase.

05

≈$500K ARR.
The region's biggest malls.

On track for ≈$800K exiting 2026 as Mall of the Emirates goes live in October.

CustomerWhat they boughtContract
Majid Al FuttaimMall of the Emirates · first of MAF's malls, all awarded$1M / 3 yrs
Galleria · AldarMall-wide analytics; client co-invested in the stack upfront$1M / 5 yrs
Aldar & MubadalaFacial-recognition security, Abu Dhabi towersvia Convergint
Mubadala · EmaarPICA self-remediation agentsper server
Retail brandsFlormar, Alaïa, Panerai and a 20-store chainper store

Dubai Festival City (Al Futtaim) is a ≈$1M deal expected next quarter; the government approval it waited on landed last week.

06

Competitors crash at a hundred cameras.
We run malls at seven hundred.

Galleria ran a competitor for four years. Reports crashed or arrived a day late, the client moved to sue, and the contract was months from cancellation. We took over, fixed the account, and the scope grew instead of dying.

600–700
cameras per mall, live reporting that holds at scale
4 yrs
of incumbent failure at Galleria before the client sued
+scope
the rescued account expanded rather than churned
07

Our second product sells itself
off the P&L.

We automated the health checks and self-remediation of our own stack because we couldn't afford engineers to babysit it. Customers saw it and demanded it. PICA now integrates with the top three security platforms and fixes tier-one and tier-two issues on the spot: no site visit, no four layers of approvals, no engineer who sleeps.

Mubadala · Emaar
first PICA customers, already paying
$100/server
20–25 servers per site under management
$1/device
edge-device tier next: a 1,000-camera site adds $1K

PICA is levels easier to sell than analytics: maintenance is a line item on every client's P&L, and saved engineer trips price the ROI by themselves.

08

The world's biggest integrator
brings us its biggest clients.

Convergint — the largest security integrator globally and in the region, now heading to IPO — brought us Aldar, Mubadala and Emaar. They like being first with something that performs, and we keep the software, the data and the client.

Convergint
sourced the enterprise book; MAF came from our own end-user push
100%
of the team local: development to sales, nobody offshore
SDRs
this round industrializes lead gen; the founder keeps closing
09

A million dollars a mall,
60–70% gross margin.

Multi-year enterprise contracts, priced on the complexity and the return of each use case. Where a client's cameras are too poor, we finance the hardware inside the deal and keep the relationship.

~$1M
average contract per mall, on 3–5 year terms
60–70%
gross margin across the book
2
products per estate: analytics and PICA on the same footprint
10

Built where the world installs
cameras fastest.

100+ new malls under Vision 2030, plus the enterprise and public infrastructure behind them.

$400M
TAM
GCC retail & mall accounts
$157M
SAM
Reachable in the 3-yr GTM
$6.4M
SOM
Target EOY 2028

Figures are retail analytics only. Security deployments and PICA run on the same estates through the same integrators, and sit outside every number above.

11

Ten years selling Gulf security.
Every client closed by the founders.

Elie Salame

CEO & Co-founder

Nine years in security tech across the GCC. Led technology at Securitas UAE, Senior Solutions Engineer at RealNetworks. Closed every client we have.

Patrick Younes

CTO & Co-founder

Previously led engineering at YUZ, Dubai. Owns the platform, the edge deployment and PICA.

Firas Al-Koussa

CISO

Eight years in cybersecurity, currently at a Fortune 500. Owns security architecture and the data-governance program.

Entirely UAE-based by design: development to sales, nobody offshore.

12

Raising $600K–$1M
to industrialize sales.

Founder-led sales won the region's biggest malls. This round adds the SDR and account-management layer under it, so lead generation scales while closing stays with the founder.

Oct '26
Mall of the Emirates live; MAF revenue starts
Q4 '26
Dubai Festival City, the next ≈$1M mall
≈$800K
ARR exiting 2026
SDR team
lead gen and account management hired with this round
OBSRVR

Recognition for the cameras
already on the wall.

≈$500K ARR across the region's biggest malls, two products on every estate, and the world's largest integrator selling alongside us.

elie.salame@obsrvr.ae
obsrvr.ae
A1

Recognition is easy
on a passport photo.

Front-facing, lit, unobstructed, one face. Every engine scores well. Nobody shops like that.

Our recognition runs on people as they actually walk through a Gulf mall: sunglasses, caps, abayas, shaylas, masks, heads turned, three-deep in a group. Engines tuned on benchmark photos degrade on exactly this.

Benchmark condition
Passport photo
Our training data
Sunglasses
Shayla & abaya
Turned away
Group, three deep
A2

Why now.

Inference collapsed

Edge recognition that cost ~$500K five years ago runs on ~$2K hardware today. Every installed camera just became a viable sensor.

The estate is being wired

100+ new malls under Vision 2030, plus the enterprise and public infrastructure behind them, contracted on multi-year terms right now.

Buyers have budgets

Mall operators and retail groups now hold AI line items, and rising shrinkage is adding a security budget alongside.