Womb Project: baby sleep and naps

The idea for this app began during my last maternity leave, from my own experience as a parent. My children's sleep was one of the most challenging aspects of that time: understanding their rhythms, anticipating naps and finding a sense of direction in everyday life. I designed the app to try to ease that mental load and help parents approach sleep with a little more peace of mind.

It is an iOS app combining a sleep journal with a personalized prediction engine. Tracking relies on local storage: manual logging and history work offline, without a required account. Optional backup and synchronization support family sharing. The app also includes voice logging, statistics and tracking for multiple children, in French and English.

The app is available on the App Store; more information on wombapp.fr.

Sleep prediction engine

To estimate the next nap, the calculation starts from the last wake-up and a reference awake duration based on the child's age and number of naps. This duration is then adjusted using recently observed patterns, prioritizing naps in the same position within the day.

Personalization uses the median of recent observations, weighted according to their number, consistency and recency. The engine also considers the previous nap's duration and usual sleep times to adjust the rest of the day. The next nap's duration is estimated separately, then used to calculate its wake-up time and plan subsequent sleep episodes through bedtime. Predictions evolve as events are logged; they of course remain guidance, rather than certain schedules.

Market validation

Baby sleep is a common concern for families, and there is already a market for paid apps that support them. The estimated revenue of competing apps such as Napper and Huckleberry shows that parents are willing to pay to track and anticipate their baby's sleep. These are encouraging signs, although they do not guarantee the commercial success of a new app.

Project technical stack

App framework and programming language

FlutterFlutter helps me develop the app and create custom interfaces. The possibility of sharing one codebase across multiple platforms is another advantage.
DartI write the app's code in Dart, the language used with Flutter, including the sleep prediction engine.

AI-assisted development

CodexCodex is my main AI development environment for implementing the app's code. With this tool, I develop features, fix bugs and write tests.
Claude CodeFor the app's UX and UI, I prefer Claude Code. At the moment, I find Anthropic's models slightly better at app design, so I turn to it to design interfaces and make visual refinements.
OpenCodeOpenCode lets me occasionally access models beyond Anthropic and OpenAI. I connect it to OpenRouter to use alternative models, particularly very low-cost ones for the simplest tasks.
OpenRouterThrough OpenCode, OpenRouter gives me access to models from different providers. I can share one budget across several models, try new ones and choose the one best suited to the task at hand.

Backend and cloud services

SupabaseWith Supabase, I manage user accounts, backups and family synchronization, with access permissions that keep each family's data separate.
CloudflareThe app's website and the Workers that power voice logging services are hosted on Cloudflare.
RevenueCatRevenueCat handles monthly and annual subscriptions, Premium access and purchase restoration.

Project management

PostHogPostHog's analytics help me understand app usage, user journeys and conversion to guide product improvements.
ObsidianI keep all the project's notes, decisions and documentation in Markdown files. This format makes collaboration with AI tools much easier.
NocoDBFor data that needs more structure than a Markdown file, I turn to NocoDB and its spreadsheet-style interface.
VoiceInkDictating text with VoiceInk instead of typing saves me a lot of time every day. VoiceInk is among the best low-cost alternatives to Wispr Flow I have tried. Its performance comes close, with the advantage of a model that runs locally.

Code management

VS CodeVS Code is the environment I use to open the project, browse files and read or edit the app's code.
GitGit lets me track code changes and isolate features and experiments with the prediction engine in branches and worktrees.
GitHubThe code is hosted on GitHub, where I can review changes and run automated checks.

Acquisition

Marketing is probably the most challenging part of this project, as I am still unfamiliar with the field. I am learning by testing different ways to introduce the app to new users. Here is what I have tried so far and what I am considering next.

Tested

Facebook account

I am testing posts in dedicated Facebook groups to introduce the app to new users. So far, the results have been limited.

Initial tests