From Lost Dreams to the App Store

He had been using a dream-journaling app and genuinely valued what was in it. The dreams got lost because the export was incomplete. The original idea was personal: build the journal he wished he had been using, where the history belongs to the user, is stored privately, and can be exported, imported, or recovered instead of vanishing with a device or a startup.

Once the journal was structured data instead of text, the larger opportunity was obvious. Someone trying to lucid dream accumulates recurring people, places, emotions, dream signs, failed reality checks, techniques that worked, and recall quality.ad Instead of the same generic advice for everyone, that history could teach a person what works for them. The journal became the foundation of a training system rather than the product itself.

The first implementation was not carefully architected. Ayden gave an agent a goal and iterated on whatever came back. That got things moving fast, and it produced the most important lesson of the project: AI can build the wrong product very quickly. Turning that experiment into a real application took roughly six weeks of rewrites.

ChatGPT Image was used to explore directions and produce a full brand kit: logo system, palette, typography, iconography, motion language, and Whisp's construction and expression sheet. Claude-driven prototyping then established screens, onboarding and interactions before the app was touched. Specifying the experience first cut the expensive downstream rewrites.

Whisp was never going to be a chatbot or a virtual pet. He is a visual guide that makes a clinical subject feel less clinical and gives the app a face. Ayden iterated unusually deeply on proportions, movement and presentation because small character details changed the personality of the whole product. The rig you see on this page is a layered character, not an image: the body, arms, face, specular light and ground shadow move independently, with an idle float, a breath, a blink cycle, and seven reactions ported value-for-value from the Swift source.

Instead of throwing advanced techniques at a new user, stage one builds recall and keeps the journal useful. The free product covers recall and journaling. Lucid Lock Pro is the adaptive training layer: a nightly practice chosen from that morning's entry, dream signs, MILD or SSILD picked from the user's own recall baseline, a review every seven practice nights, and Insights. The principle that came out of it: recall is free, the program is the subscription.

The original ambition was to analyze dreams with Apple's on-device intelligence and generate personal observations. In practice the small models were inconsistent at reading entries. Ayden tested generated Whisp notes and coaching, then chose not to make unreliable generative analysis load-bearing. Production uses deterministic pattern matching and structured signals first, with NaturalLanguage embeddings for recurring situations and Apple Intelligence only phrasing results where present. The generated note stays behind a flag that is off in release. It's a mature call: he removed impressive technology when it didn't reliably improve the product.

Privacy and data ownership came directly from the event that caused the app. Local persistence, export and import, Face ID, a full delete, and an iCloud sync that only ever goes to the user's own account. There is no service whose usefulness depends on Ayden's server staying online, because there is no server.

Testing exposed what Ayden's own use never could. On the onboarding subscription screen, the chevron he understood as an exit read as ordinary back navigation to someone else, so she concluded the app required payment simply to keep a journal. That misreading led to naming the paid layer Lucid Lock Pro, a visible labelled decline button, an explicit screen saying the journal is unlimited and free forever, and removing promises about how quickly anyone becomes lucid, because the app cannot honestly guarantee it.

The higher monthly number created too much friction against comparable consumer wellness apps. A conventional trial was rejected on purpose: users experience the recall system for free and meet Pro only when they reach functionality that actually requires the training system. That is a more defensible paywall than locking the journal after a few days.

The listing got the same treatment as the product. Sets A through F each have a written plan and a build script. Set A opened on a milestone ring that said nothing at thumbnail size. Set F opens on a real tester quote with written permission behind it, states the privacy claim in a form a competitor can't copy, and reads in order as a course. Frame 4 is composited from two physical-device captures because the pattern engine won't run on a simulator.

By late August the app was in good shape and almost nobody could find it. Instagram, TikTok, X and YouTube experiments ran in parallel: dream phenomena, relatable dream problems, lucid-dream education, static versus video. A new Instagram account was permanently banned. TikTok results swung wildly post to post. Atlanta Tech Week showed the subject has real pull, with people pulling Ayden aside to talk about lucid dreaming and early testers coming out of those conversations. An existing YouTube audience built on Mario and game development did not transfer. Building a product and distributing one turned out to be almost separate disciplines. The one channel that is fully under his control is the website, lucidlock.co, which carries the same privacy claims as the listing and links straight to the App Store.

Subscriptions, App Store Connect agreements, forming PixelNova LLC, converting the developer account into an organization, D-U-N-S, banking, tax, a business address, a support inbox, screenshots, metadata, TestFlight, and protecting production data while continuing to test. Ayden was using Lucid Lock as his own dream journal the whole time, so destructive testing could destroy data he actually cared about. The app was technically finished before every administrative dependency was resolved.

After the metadata fix, Apple reviewed the build quickly and approved it without a prolonged rejection cycle. It moved to Ready for Distribution. The project is no longer an app being built. It's a product a stranger can install, and the question it can now answer is the one nothing before launch could: whether people stay. It's live at lucidlock.co and on the App Store.