← PERCEPTA
For investors

Citizens, government and industry each see a different version of the same city.

PERCEPTA turns the street cameras a city already owns into one shared civic dataset — published to all three at once. Avanti R&D has built vision systems for automotive, industrial and government clients since 2016, and we own all of the IP in the platform, unencumbered.

The problem

Dormant camera assets Governments already operate thousands of CCTV cameras, yet they are used only for after-the-fact monitoring. The video creates no value.
Reactive city management Accident statistics are lagging indicators — visible only after harm is done. Without leading indicators like near-misses, a city cannot invest in prevention.
Citizen voices don't get through Phone-and-counter reporting means vague locations, duplicates and privacy risk. The information asymmetry between residents and government persists.

The solution

Zero new cameras. Existing CCTV becomes a shared data platform, and raw video never leaves the premises.

1
Existing CCTVGovernment-managed cameras connected as they are. No new hardware.
2
On-premises AIDetection, plus face and plate anonymisation, on site. Raw video never leaves.
3
Cloud aggregationOnly detection events and statistics are sent. Low-cost by construction.
4
Three-tier publishingCitizen dashboards · industry APIs · government console.

And a fourth source flows back in. Citizen app reports enter the same platform, where camera detections and resident reports cross-verify each other.

Why now

01
The cameras are already there Cities operate thousands of government CCTV cameras. For the first time, citywide coverage is possible with zero hardware investment.
02
Edge AI has matured Detection and anonymisation now run entirely on-premises and on-device. Privacy regulation and video analytics are finally technically compatible.
03
Southeast Asia tailwinds Thailand's PDPA is in force and smart-city programmes are accelerating. No PDPA-compliant reference exists yet — the first-mover slot is open.

Why we are productising

Why the model has to change · illustrative
Services revenue
One large programme lands, and the next year collapses. Every dollar of it has to be won again from nothing.
Per-camera recurring
A city renews. A data licence accrues. Last year's work is still earning while this year's is being signed.

The shape on the left is why we are doing this. A decade of delivered engineering proved the capability, but services revenue never builds on itself — each year starts again from zero. PERCEPTA sells the same detection stack as a subscription a city renews, plus a data layer that scales with connected cameras.

No figures here by design. The real ones are the first thing we will take you through under NDA.

What defends it

A three-sided data market The same anonymised events published to residents, to government departments and to industry under licence. Every side added makes the data worth more to the other two — more cameras widen what residents and industry can see, and more residents reporting fills in what no camera can resolve.
Revenue share back to the city A model that turns infrastructure a city has already paid for into something that contributes to its own running cost, rather than only consuming budget.
Cross-verification Cameras verify residents; residents verify cameras. A pothole reported by three people and seen by a camera is a work order; one without the other is a lead. It spans both halves of the problem.
The permit, already cleared Putting AI onto a city's camera network requires an agreement with that city, and those run for months to years. A government MOU plus a Test Bed on government-managed CCTV means a customer skips that process entirely and starts validating immediately.
Privacy by construction Blur is applied before the encoder, so personal data is never created rather than protected after the fact — and every delivered row carries its own provenance proving it. The right to license civic data rests on anonymisation being demonstrable, not merely true.

The market

TAM
$28.8B
Global smart-city AI video analytics and urban data, 2030 — from $8.5B in 2025, 27.6% CAGR
SAM
$2.8B
ASEAN metros × existing-CCTV analytics and urban data APIs, 2030 estimate
SOM
$13M
First three years: two Thai cities into major ASEAN metros — 26 ASCN cities at mature ACV

TAM: MarketsandMarkets, AI Video Analytics for Smart Cities (June 2026). SAM: smart-city video analytics and urban-data share of the ASEAN AI computer-vision market (DataCube). SOM: 26 ASEAN Smart Cities Network cities at our modelled mature ACV. Full detail and sensitivities in the data room.

Where we are

Built and running

  • The platform is in production — running today on a city's real cameras, with a public dashboard anyone can open
  • Phuket is signed for citywide deployment — the direct agreement, with full freedom over data publishing and PoC design
  • A Test Bed on government-managed CCTV — the sales asset that lets a customer skip the permit process
  • The data API is built — authentication, per-city scoping, metering, 19 event types, public documentation
  • Citizen app validated on real Android hardware, masking faces and plates on-device in about four seconds
  • Europe since 2018 — Copenhagen, the DOLL Living Lab in Denmark through 2025, and autonomous-shuttle perception in Estonia with a completed GDPR DPIA

What comes next

  • The Phuket PoC. Ten cameras and the core features, with quantified KPIs
  • Free pilot to paid deployment. The conversion that turns a reference into revenue
  • The per-camera pricing model, proven. Several paying cities on the same terms
  • The 1,000-camera commercial architecture validated — the gate to a Series A

Who acquires this, and why

Strategic buyers appear unusually early in this category, and three of the four groups below are already in discussions with us at seed stage.

Automotive OEMsReal-world validation environments and city data for ADAS and autonomy — plus government access they cannot build in-house.
Tier-1 suppliersStronger sensing and perception algorithms, and a secured infrastructure-side data source.
Camera & chip makersThe software layer their hardware needs, and a way to monetise installed CCTV assets.
Mapping, MaaS & insuranceFresher map data, and risk scoring built on near-miss data that exists nowhere else.
The moat itself is acquirable Government agreements plus an existing-CCTV Test Bed. An acquirer inherits a permitting process that normally takes months to years.
Not one-off — continuously supplied Not a technology that ships and ends, but a data source that stays connected to the city and compounds after acquisition.
Faster than building it Earning a city's trust typically takes years. This is a domain where buying genuinely beats building.

Go and check

Barcelona · demonstration
Open the running platform →
Not a slide. The real system, showing what Barcelona would look like on it; the same build runs live today on a city's own cameras. No sign-in.

If this is your area

See it before the round opens →
We open our seed round in Q4 2026 and are not raising on the exhibition floor. Leave your details and you will see the materials before they go out broadly. Terms are not set, and we would rather not pretend otherwise on a web page.