Same foundation. Second app. Shipped faster.

The big weather apps all do roughly the same thing, and most of them bury it under ads, trackers and features nobody asked for. Stormline started as a reaction to that: a native weather app built around what matters — a clean forecast on one scrolling screen — plus the two things that are actually worth paying attention to: plain-language AI summaries of what the weather is about to do, and alerts for lightning and severe weather nearby.

No ads, no advertising identifiers, and no third-party analytics or crash-reporting tools. That's not a missing feature; it's the point.

Stormline began life as a personal weather app. Turning it into a public product meant rebuilding it properly: separate iPhone, Android and backend projects, each held to the same build, release, diagnostics and logging standards that Scavenger had already proven in production.

The Android app is the clearest example. Its very first commit already had the release build hardening that Scavenger only added after a security review, the groundwork for its own built-in crash reporting (no third-party crash SDK), and an automated pipeline where tagging a release builds it and delivers it to Google Play's testing track — the same release process Scavenger uses, carried over almost line for line. The code that checks App Store purchases was ported straight from Scavenger's production backend rather than written again.

Lessons carried over too, not just code. Problems Scavenger had already hit and fixed — a purchase screen that silently failed to appear, a sign-in setup step that is easy to miss, a shared AI key that let one app's trouble affect the other — were designed out of Stormline before they could happen. Each app now has its own AI key, so a problem with one can't take down the other.

AI tools did much of the heavy lifting on speed; experienced engineering judgment decided what got built, and checked what actually shipped. The result: about a week from the first Android commit to the 1.0 release build.

The personal version of the app pulled forecasts from four different commercial weather providers. Stormline replaced all of them with one: a forecast service that comes included with a developer membership we already pay for, up to a generous monthly allowance — so there's no per-call licensing bill growing alongside the user base. The other data the app uses comes from public sources: severe-weather alerts from the US National Weather Service and Environment Canada, radar from Environment Canada, and lightning from NOAA's weather satellites.

Each piece runs where it makes sense. The app's API and its caching run on commercial-grade cloud hosting at the network edge, close to users and cheap to run, with forecasts cached briefly so the same neighborhood isn't fetched twice in a row. Lightning detection is a different kind of job — regularly downloading and processing satellite data files — so it runs as a small, separate cloud service that does only that, and only does work for places someone has actually asked to watch.

Being a good citizen of free public data is part of keeping it cheap. The app only fetches radar when someone is looking at it, never crawls or pre-loads it, and has hard daily limits on how often it asks — so it stays well within each provider's usage rules.

Stormline works fully without an account. Forecasts, radar, and lightning and severe-weather alerts all work for someone who never signs in. Signing in with Apple or Google is only needed for two things: syncing places between devices, and AI summaries, which need a way to track a free allowance and a subscription.

Even then, the account holds as little as possible — an anonymous ID from Apple or Google, not a name or email address. Alert registrations are tied to the device, not the account, and expire on their own if the app stops checking in. Deleting an account takes a few taps in the iPhone app or a short form on the web. Collecting less data means there's less to protect, less to leak, and less to explain.

AI summaries are the paid feature, under a single Stormline Weather Pro subscription. New users who sign in get 10 free days of summaries to try it — a "day" being one place on one date, however often it's opened — and the annual plan adds a free trial through the app store.

Behind that sits the same proportionate thinking as Scavenger. Every AI request costs real money, so there are limits at every level: per person per day, and a ceiling for the whole service. Those limits can be changed live without an app update. If something does go wrong, AI summaries or the free trial can be paused on just one platform rather than switched off everywhere, and a full kill switch exists for the worst case. Repeat views of a summary are served from a cache at no cost, and a failed summary is never counted against anyone's allowance.

Those controls, and the alerting that watches both apps, live in the Northbaseworks Control Center, the operations console that started with Scavenger and now runs every app we ship.

Android went first, and Stormline is now live on Google Play.Live

The iPhone and iPad app has been submitted to Apple and is in App Review.In review As with every Northbaseworks case study, that status is what's true as of this writing, not what's hoped for.

Launch wasn't the finish line. Because releases are a single tagged build away, the first days after the Android launch brought a quick run of new builds driven by real use: radar refinements, a home-screen widget whose colors follow the weather and time of day, and alerts that open straight to the alert you tapped.