A movie deck for the next decision
Give the next film a moment.
Artwork sets the scene. A title, runtime, ratings and a short description help turn a passing suggestion into a decision. The trailer stays within reach.
01 / 03I wanted choosing a film to feel like a small, manageable decision. KinoCue presents movies as a deck: look at a card, explore the details, save something for later, mark it seen or move on. A watchlist gives those choices somewhere to live after the browsing session ends.
Filters narrow the search by things such as genre, release year, runtime and rating. An optional voice note lets you describe what you feel like watching. The app turns that request into preferences that can shape the next recommendations.
KinoCue is a native SwiftUI iPhone app. Its saved profile, watchlist and history live on the device; movie discovery and voice processing use backend services. The experience centers on choosing a film, with movie information and embedded trailers helping you make that choice.
An iOS workflow inside my Windows workspace
01WindowsDirection & review
The workbench holds the request, the feedback and the next decision. I can review the app while staying in my Windows workspace.
02MacBuild & run
Xcode owns the build and the Simulator. The development workflow connects the request to the machine that can actually run the iOS app.
03iPhone SimulatorObserve & refine
The app launches in the Simulator. Its rendered interface and interactions give the next round of feedback something concrete to respond to.
An explanation of the development setup. These controls do not connect to a computer or run tools.
I worked on KinoCue with Codex from Windows. Source inspection, changes and the development conversation happened in that workspace, while a connected Mac ran Apple's toolchain. Xcode still built the app, and the iOS Simulator still ran on the Mac.
The useful part was bringing the live Simulator into the Windows workspace. I could keep the implementation, the agent's work and the running interface close together, then judge the result through the app itself. The bridge made the remote Mac part of a usable development loop.
That loop also needed a reliable handoff. A passing Mac test only says something about a Windows edit if the same source reached the build host. Keeping source transfer, the generated Xcode workspace and the tested binary aligned became part of the work. I described this setup in my development post on LinkedIn.
Finding the right interaction
The visual work went beyond the movie card. A dark, poster-led deck loses its coherence if opening the filters or profile suddenly feels like entering another app. I wanted those screens, the voice composer and the introduction to share the same surfaces, contrast and visual language.
The implementation grew a shared theme across those views. Alongside appearance, the saved-film flow became more explicit: find a movie, put it on the watchlist, return to it and mark it watched. The introduction needed to lead into those real actions.
That gave the project a clearer purpose. The card is the beginning of a decision; the watchlist lets you carry it forward.
Making the details earn their place
Testing the app on an iPhone caught problems that were easy to miss in a tidy walkthrough. A learning hint appeared again on successive cards while I swiped. A voice request for animated films with very high IMDb ratings produced a deck that did not respect the requested genre.
I reported both from using the app. The fixes made the hint a one-time event and turned the specific voice request into enforceable constraints: animation, with an IMDb minimum of 8. A general preference for animation could still behave as a preference without silently adding that rating threshold.
The correction crossed the app and its backend, so a successful screen change alone was insufficient. The work included testing the interpreted request, the resulting filters and the updated app on an iPhone Air. Those small failures were useful: they showed exactly where the interface's promise and its behavior had come apart.
The road through App Review
An earlier App Review rejection under Guideline 4.2.2 questioned the app's standalone functionality. That pushed the work toward a more complete personal workflow: a discoverable watchlist, persistent choices, watched status and an introduction that helps someone use them.
The release journey continued through replacement builds and another round of fixes. The dated delivery milestone recorded here is KinoCue 1.0, build 7, verified as Waiting for Review on 6 September 2026. Automatic release after approval remained configured. Public App Store availability has not been confirmed for this case.
What KinoCue taught me
KinoCue brought product direction and development infrastructure into the same project. I had to make the app useful, make its interface coherent and make a Windows-to-Mac workflow trustworthy enough to support repeated testing.
My part was to set the direction, use the result, notice where it failed and keep refining the experience with Codex. The strongest progress came from a concrete observation: a screen that felt out of place, a hint that would not go away, or a spoken request that the next deck did not honor.