31/08/2026
When we first mapped out our hackathon project, our vision was massive. We set out to build an AI-powered missing person recognition system for Cambodiaβan edge-camera pipeline that extracted face embeddings to replace the slow, manual photo comparisons happening across local stations.
The technical blueprint was solid, but as we moved into customer discovery, the reality of a B2G/B2NGO model hit us hard. The real hurdle wasn't the AI; it was the red tape, the consent records, and the years it takes to get government institutions to adopt a system built by young founders.
We didn't have years. We had days.
So, we made the hardest choice a startup team can make: we killed our darling and pivoted.
We took a step back, looked at the core technology we had just builtβreal-time facial scanning, edge-compute logic, and instant Telegram alertsβand asked ourselves: Who else in Phnom Penh needs a localized, automated set of eyes right now, with zero red tape?
To find out, we closed our laptops and hit the streets. We sat down for coffees with independent cafe owners, talked to retail managers, and even met with a local high-end automotive tuning garage. We didn't pitch them code; we just listened to their daily headaches.
What we found wasn't just a single issueβit was a massive, fragmented market. The garage wanted to track how much time mechanics spent inside specific work zones. The cafes wanted automated staff attendance. Retail managers wanted foot-traffic validation. Every business owner had their own unique blind spot, but the underlying truth was crystal clear: standard CCTV is dead, and merchants desperately need their cameras to actually work for them.
We knew the demand for a multi-feature AI surveillance platform was huge. But to build a real startup, you can't boil the ocean on day one. You need a wedge. One problem. One solution.
To grab the market and prove our tech, we had to start with something highly believable and completely frictionless.
So, we stripped it all back to the most universal, painful problem across the board: the unattended front counter. No complex multi-zone enterprise setups yet. Just our lightweight software running silently on their existing cashier PC, instantly pinging the owner's phone with a Telegram photo the exact second a customer is left waiting.
This is what building actually looks like. Itβs looking a stressed business owner in the eye, seeing a massive market potential, but having the strategic discipline to start with just one believable solution.
The core mission hasn't changedβwe are still using AI to give people their peace of mind back. We just found our wedge. π
Β°
Β°
Β°
