The Story So Far, 727 Commits Later

This is post two, and it's the one I skipped by starting with launch day instead of the beginning. Here's the actual timeline, as honestly as the commit history and session logs can tell it.

Before WhatNext., There Was Pols

WhatNext. isn't the first app. Pols is, a workout tracker I started building in October 2025, about seven months before WhatNext.'s first commit. It has its own 459 commits. It never went anywhere near the App Store. And it's still the app I actually open at the gym, most days, right now, while WhatNext. is live and Pols isn't.

That's not really a failure story. Some things get built because you need them, not because you're trying to launch a company. Pols does exactly what I built it to do, so there was never a real reason to push it further than that. WhatNext. started from a different place: something I wanted to exist for other people, not just for me, which is most of why one of these apps has an App Store listing and the other has a folder on my laptop.

727 Commits Later

WhatNext.'s first commit was a product spec and a database schema, written before a single screen existed. From there to launch was about three and a half months and, as of this post, 727 commits: pipeline work, algorithm iteration, a full catalogue quality pass, the onboarding flow, four separate trips through Apple's review queue, and, more recently, this blog.

That number doesn't mean much on its own. What it actually represents is the boring middle of building something solo: most commits aren't features, they're fixes to things the last commit broke, or data cleanup, or a rewritten prompt that improved one specific case by a few percentage points. The recommendation engine's current cold-start behavior, the one that held up an already-approved build, is itself the product of a dozen smaller commits arguing with each other before the real fix landed.

Four Trips to the Review Queue

The formal rejection is the easy one to tell: Guideline 2.1(a), "the app became unresponsive after onboarding." The reviewer's screenshot showed them stuck on the tutorial screen. Root cause found, fixed, and resubmitted within the week.

The more interesting one is the build that never got rejected at all. Build 11 went to Apple on June 11 and came back approved. Under a normal plan, that ships, and a cold-start bug (new users seeing visibly worse recommendations for roughly their first ten ratings, before their taste profile has enough signal to work with) gets patched as a fast-follow update a few days later. That was the actual plan, briefly. Then it changed: hold the approved build, fix the cold-start behavior properly, and resubmit a new one, which meant giving up an approval already earned and going back through Apple's review process from the start.

There were earlier rounds too, pulling a submission back before Apple had even finished reviewing it, mostly to fix things a self-run content or metadata audit caught first, rather than waiting for Apple to catch them. None of those were formal rejections. They were closer to catching your own mistake before someone else does, which is a less dramatic story but probably the more honest description of most of what "getting an app through App Review" actually looks like.

What's Next

WhatNext. is live now, so the story from here is less about getting to launch and more about what happens after. That's most of what this blog will be: the mechanics of the recommendation engine, decisions that looked obvious in hindsight and weren't at the time, and, since it's still a movie app, actual recommendations along the way.