Android App Development Company | Kotlin, Jetpack Compose, Android 17
Android app development in 2026 means Kotlin and Jetpack Compose, a layout that resizes correctly on every screen, and a release plan built around Google Play's target-API deadlines. Mixcore Studio builds new Android apps and takes over existing ones, with the compliance work that now sits between a finished build and a published update treated as part of the engagement rather than an afterthought.
We have been shipping software for more than 8 years, with a team of around 20 and over 320 projects delivered across web, mobile and enterprise systems. On Android that experience mostly shows up in the unglamorous parts: build infrastructure that survives an annual toolchain bump, layouts that were never hardcoded to a phone in the first place, and release pipelines that do not stall the week a policy deadline lands.
What changed in Android during 2026
Android UI development is now officially Compose-first, declared at Google I/O 2026. According to the Android Developers Blog post "Celebrating 5 years of Jetpack Compose" (28 July 2026), more than 68% of the top 1,000 Android apps use Jetpack Compose in production. The same post puts the Views toolkit into maintenance mode and states that Material Design is shifting focus entirely to Compose.
Android 17, API level 37, reached stable in June 2026. It introduces a permission-free system contacts picker, OTP protection for standard SMS so non-priority apps no longer get immediate access to one-time passwords, post-quantum APK signing, Certificate Transparency on by default for apps targeting 37 and above, a cross-device Handoff API, WebGPU, and a lock-free MessageQueue implementation for apps targeting SDK 37 and above.
Several long-standing practices stopped being viable this year. If your codebase still relies on any of these, the work is scheduled rather than optional.
- Opting out of resizability and orientation constraints — no longer possible on large screens, meaning a smallest width of 600dp and above, in Android 17, so letterboxing on tablets and foldables is not a supported shipping state.
- Broad READ_CONTACTS for contact selection — replaced by the system Contact Picker, with the same pattern already established for images via the system photo picker.
- ExoPlayer 2 — the standalone com.google.android.exoplayer2 library is discontinued and AndroidX Media3, at 1.11 as of 11 August 2026, is its stated home. Separately, Android 17 hardens background audio: apps targeting it need a running foreground service with while-in-use capability to touch audio in the background.
- The Wear OS Tiles framework — superseded by Wear Widgets built on Jetpack Glance and RemoteCompose in Wear OS 7.
- Modifier.onFirstVisible() — deprecated in the August 2026 Compose release in favour of Modifier.onVisibilityChanged(), which lands alongside new test APIs such as hasPendingWork() and runWithoutImplicitWait() for cutting flakiness.
- Unrestricted local-network access and reflection on static final fields — the first now requires the ACCESS_LOCAL_NETWORK runtime permission, the second is blocked outright on Android 17.
The Google Play deadlines that block an update
These are hard cut-offs that stop a release from being accepted, not recommendations. Google Play Console Help states that new apps and updates must target API 36 (Android 16) by 31 August 2026, with an extension available to 1 November 2026, and that existing apps must target at least API 35 to remain available to new users on newer devices. Wear OS and Android Automotive require API 35 or higher; Android TV and Android XR require API 34 or higher. The required level moves up as each new Android version ships, so this is now an annual cycle rather than a one-off.
- Developer verification — the developer APIs launched in August 2026 and enforcement begins 30 September 2026 in Brazil, Indonesia, Singapore and Thailand, covering Google Play, HONOR App Market, OPPO App Market, Galaxy Store, Palm Store, V-Appstore and GetApps on certified Android 7+ devices. Global rollout follows from 2027. An advanced flow preserves sideloading of unverified apps for experienced users, which matters for internal and enterprise distribution.
- Age signals — the Play Age Signals API returns an age range shared through Family Link. It is already live in Brazil and arrives in Australia and Canada from mid-August 2026, with global availability later in the year. Play Console also offers Restrict Minor Access controls.
- Regulatory pressure behind it — US state app-store accountability laws now push the job of knowing whether a user is a minor onto the store and the app. The practical consequence for developers is consuming the store's age signal rather than self-attesting.
- Billing options — Play policy states that a US District Court order remains in effect allowing US developers to offer alternative billing options or external redirects, and that alternative billing programmes exist for the EEA, India and South Korea, with a 4% service-fee reduction in India and South Korea. Antitrust enforcement against Google's app-store and steering rules is still live on both sides of the Atlantic, so treat the specifics as a moving target at contract time.
On billing we give a recommendation rather than a survey. The number worth optimising is net revenue after engineering, payment processing, tax handling, refunds and support, which is frequently worse than the headline commission suggests once a second payment path has to be built and staffed.
The stack we build on
Kotlin 2.4.10 is the current stable release, from 14 July 2026, with 2.4.20 planned for September 2026 and 2.5.0 for December 2026. Jetpack Compose 1.12 and BOM 2026.08.00 shipped on 12 August 2026 alongside material3 1.4.0 and material3-adaptive 1.3.0. Our default tooling is Android Studio Quail 3 (2026.1.3 Patch 1), released 28 April 2026.
- Compose 1.12 additions we use in production — MeshGradientPainter, wide-colour P3 and HDR support, DeferredAnimatedContent and DeferredAnimatedVisibility for gesture-driven two-stage transitions, rich-text formatting in BasicTextField via TextFieldBuffer.addStyle(), programmatic SelectionState, a credentialRequest semantics property for Credential Manager and passkeys, and named areas in the experimental Grid.
- Measured performance work — the Android Developers Blog notes in "What's new in the Jetpack Compose August '26 release" (12 August 2026) that a keyed SideEffect is up to 90% faster than LaunchedEffect and around 20% faster than DisposableEffect, which is the kind of change worth applying to hot recomposition paths rather than everywhere.
- A build-infrastructure cost to budget — Compose 1.12 requires compileSdk 37 and Android Gradle Plugin 9.1.1 or newer. On an older project that is a Gradle and CI upgrade before it is a UI upgrade, and it deserves its own line in the estimate.
Adaptive layouts, foldables and Wear OS
Android 17 removed the escape hatch, so any app with hardcoded phone layouts now has mandatory work above 600dp. The commercial argument is stronger than the compliance one: the Android Developers Blog reported on 12 August 2026 that foldable users spend about 14 times more than standard phone users.
We build against Window Size Classes and the material3-adaptive library rather than device checks, which keeps the same layout logic working on tablets, foldables, desktop windowing and free-form resizing.
Wear OS 7 shipped with the Pixel Watch 5, adding one-handed gestures such as double-pinch for a primary action and wrist turn for dismiss, and Wear Widgets built on Jetpack Glance and RemoteCompose in place of the old Tiles framework. The Android Developers Blog reports up to 10% better battery life than Wear OS 6. Compose for Wear OS 1.7 beta adds native gesture integration. If you already have a Tiles implementation, treat it as a rewrite rather than a port.
On-device AI with Gemini Nano 4
Gemini Nano 4 is reachable from application code through the ML Kit GenAI Prompt API and AICore, with structured output and a thinking mode for harder reasoning. It ships on the Pixel 11 family, and the Android Developers Blog reports that the Prompt API covers over 140 languages. Alongside Prompt, ML Kit GenAI exposes ready-made summarisation, proofreading, rewriting, image description and speech recognition APIs over the same on-device model, which is usually the cheaper starting point than driving AICore directly.
The honest framing is a split rather than a wholesale move. On-device models are well suited to summarisation, extraction, classification, rewriting and structured output, and they are free at the point of use, work offline and keep data on the handset. They are not a substitute for world knowledge or hard multi-step reasoning, and a meaningful share of most install bases will never run on a device that supports them. We design the feature with a cloud fallback and an inference budget from the start, and we tell you up front what fraction of your users are likely to see the local path. If the same feature also ships on iOS, note that App Review Guideline 5.1.2(i) requires explicit user permission and clear disclosure before personal data is shared with any third-party AI, which turns the work into a consent-flow and vendor-review project as well as an engineering one.
When native Android is the wrong choice
Two native codebases are hard to justify for a content-driven or form-driven product with one small team and feature parity expected on both platforms. In that case the current default is to share logic and keep native UI, which in practice means Kotlin Multiplatform. KMP is production-standard at Google Workspace, Duolingo, Forbes and Philips. The JetBrains blog post "Helping Decision-Makers Say Yes to KMP" (April 2026) reports that Duolingo saved an estimated 6 to 12 engineer-months, spending five months on iOS and 1.5 months on web against nine months for the initial Android build, that Bitkey shares 95% of its mobile codebase, and that Forbes consolidated over 80% of its logic across platforms. Compose Multiplatform 1.11.0 (May 2026) added UIView-backed native iOS text input and turned on concurrent iOS rendering by default. Flutter and React Native remain reasonable choices for other shapes of product, and we cover that trade-off on our cross-platform page.
Native Android is the right call when the app depends on deep platform integration: background media, camera and sensor pipelines, on-device ML, tight startup and frame budgets, or presence on Wear OS, Android TV, Automotive and XR. It is also the right call when the Android app is the product rather than one of several clients.
A related judgement is whether to migrate to Compose at all. For a large, stable View-based app with no active feature roadmap, a wholesale rewrite is rarely worth funding. Compose interop lets you convert one screen at a time as those screens are touched anyway, and we would rather scope that than sell a rewrite that buys you nothing you can measure.
Because both the toolchain and the target SDK now move on an annual schedule, we usually propose a named maintenance retainer covering the yearly target-API bump, dependency upgrades and policy changes, alongside any fixed-price build. That is a more honest structure than a fixed-price project that quietly stops being publishable a year later.
Our expertise
- Kotlin and Jetpack Compose
- Adaptive and foldable layouts
- Play policy and release compliance
- On-device AI with Gemini Nano
- Wear OS and large-screen surfaces
- Startup time and runtime performance
Frequently asked questions
Do we have to rewrite our Android app in Jetpack Compose?
Not wholesale. Compose and the Views framework interoperate, so the usual path is to convert screens as they are being changed for other reasons rather than funding a rewrite on its own. Compose is the default for new work, and the Android Developers Blog reported in July 2026 that more than 68% of the top 1,000 Android apps now use it in production, but a stable View-based app with no feature roadmap is a poor candidate for migration.
What is the deadline for targeting API 36 on Google Play?
Google Play Console Help states that new apps and updates must target API 36 (Android 16) by 31 August 2026, with an extension available to 1 November 2026. Existing apps must target at least API 35 to stay available to new users on newer devices. Wear OS and Android Automotive need API 35 or higher, Android TV and Android XR need API 34 or higher. The required level then moves up as each new Android version ships, so plan for it annually.
What is Android developer verification and does it affect our app?
Enforcement begins on 30 September 2026 in Brazil, Indonesia, Singapore and Thailand, covering Google Play, HONOR App Market, OPPO App Market, Galaxy Store, Palm Store, V-Appstore and GetApps on certified Android 7 and newer devices, with global rollout from 2027. It matters most if you distribute outside Play in those markets, including enterprise and internal builds. An advanced flow still allows experienced users to install unverified apps.
Should we build native Android or use a cross-platform framework?
Build native when the app depends on background media, camera or sensor pipelines, on-device ML, tight performance budgets, or Wear OS, TV, Automotive and XR surfaces. Share code when the product is content or form driven and one team must ship both platforms. Kotlin Multiplatform is usually the lower-risk version of sharing, since the UI stays native. JetBrains reports Duolingo saved an estimated 6 to 12 engineer-months with KMP and Bitkey shares 95% of its mobile codebase.
Can we run AI features on-device on Android, and what are the limits?
Yes. Gemini Nano 4 is callable through the ML Kit GenAI Prompt API and AICore, and the Android Developers Blog reports the Prompt API supports over 140 languages. ML Kit GenAI also ships ready-made summarisation, proofreading, rewriting, image description and speech recognition APIs over the same model. The limits are real: on-device models suit summarisation, extraction, classification and structured output rather than world knowledge or hard reasoning, and older devices in your install base will need a cloud fallback that carries an inference bill.
Contacts
We are always happy to talk with you.
Feel free to contact us in any suitable way
Request a quote
Let's discuss your project!
Please, provide us with a brief description of what you
already have and what you are going to achieve.
Mail us contact@brainiacminds.com