Clixbee

Clixbee started with a small, familiar moment. My daughter was scrolling through a group of nearly identical photos, trying to decide which one to send to friends. None were clearly bad. None were clearly better. She flipped between them, zoomed in, hesitated, and eventually guessed. The camera roll treats every photo the same, even though people do not.

Around the same time, AI-assisted coding tools made it practical for me to build apps hands on again. I had shipped iOS apps earlier in my career, then spent years leading product and platform work. The new tooling lowered the friction enough to return to solo development and ship something end to end.

The app answers a simple question: which photo is the best? It evaluates images on device and surfaces the strongest option within a set. Clixbee automatically analyzes recent photos or takes a user-selected group of phoros, then scores them.

Under the hood, Clixbee runs on a modular scoring system. Each dimension is evaluated independently so composition, exposure, clarity, and other signals can evolve without rewriting the entire engine. Scoring also runs on device to balance performance and privacy.

Building Clixbee has reinforced several patterns I have seen elsewhere. AI systems are most helpful when they narrow decisions rather than expand them, and modular architectures make it easier to improve small parts without destabilizing the whole product. These new AI tools have also changed what a single builder can reasonably ship.