WizMark

WizMark app icon — original, uncropped 1024 × 1024.
Home — collections and social shortcuts. Original 1179 × 2556 capture; sample data.
Collections keep saved links recognizable with thumbnails, source domains and manual tags. Sample data.
AI summary and separate place details from one saved article. Prefilled sample results; not a live extraction.
WizMark Pro — AI features, data export and ad removal, with a one-week trial for eligible new subscribers.

Inspiration

Saving a link is easy. Remembering why it mattered—and finding the useful details again—is harder. A restaurant recommendation, a recipe, and an article can all end up in the same growing pile of links. Opening every page again turns a small decision into another research session.

WizMark starts with that everyday problem: a bookmark should preserve the context that made something worth saving. The goal is to make saved links useful when it is time to choose a place, revisit an idea, or plan something with another person.

What it does

WizMark is a native iPhone and iPad app for saving, organizing, and revisiting links. Send a URL through the iOS share sheet, and WizMark fetches page metadata for a recognizable bookmark. Add your own notes and tags, organize links into collections, and search across saved content.

WizMark Pro adds AI enrichment. Users choose the information they want, including summaries and place details. A page recommending several cafés can become a bookmark with separate place names, addresses, and opening hours. Places can open in a maps app. Collection settings can remember extraction preferences, and AI can suggest an existing collection for an unfiled bookmark.

Shared collections let people gather links together, with viewer and editor permissions. Basic personal bookmarking works without creating an account. Personal bookmarks are stored locally and can sync through iCloud; shared collections use a separate authenticated service.

How we built it

The app uses Swift 6 and SwiftUI, with SwiftData for local persistence and CloudKit for personal synchronization. A native Share Extension keeps saving close to the app where a link was discovered. MapKit connects saved place information with a visual map.

AI extraction runs through a Convex backend using Gemini 2.5 Flash with Google Search grounding. Clerk supports sign-in for shared features. The interface displays analysis status so a saved bookmark remains understandable while enrichment is pending or has failed.

RevenueCat handles subscription offerings, purchase and restore flows, and entitlement updates. The pro entitlement enables AI features, an ad-free experience, and data export. A separate ad_free entitlement supports a one-time purchase for people who only want to remove ads. Purchase status also reaches the Share Extension, keeping premium access consistent while saving.

Challenges we ran into

A bookmarked page does not always describe one thing. An article can recommend several places, and their names, addresses, and opening hours must stay together. WizMark represents those places as grouped records, preserving the information for each destination.

Another challenge is keeping saving dependable while network-dependent work happens separately. The bookmark and its enrichment have distinct states: a failed AI request should not erase the link the user wanted to keep. Coordinating local data, personal sync, collaboration, and purchase state also required clear boundaries between services.

Accomplishments that we're proud of

WizMark shipped on the App Store on August 2, 2026. It brings a native sharing workflow, local storage, AI enrichment, collaborative collections, and RevenueCat monetization into a single app available in English and Japanese. The result goes beyond storing URLs: a saved recommendation can become something a person can inspect, annotate, share, and act on.

What we learned

Useful AI needs a clear place in an ordinary workflow. Choosing what to extract, showing progress, preserving the original link, and making results easy to copy matter alongside the model itself. Separating Pro from ad removal also makes the product's paid choices easier to explain.

What's next for WizMark

The next priorities are to refine extraction quality across more kinds of pages, improve the path from saving to revisiting, and learn from real usage of collaborative collections. The guiding question remains simple: does this help someone do more with a link they already thought was worth keeping?