RunAnywhere AI — Android Example

A production-ready reference app for the RunAnywhere Kotlin SDK. Chat, speech, vision, voice agents, RAG, and model management—all running on-device with privacy-first, offline-capable inference.

---

Requirements

Item	Minimum
Android Studio	Latest stable (Ladybug or newer recommended)
Android SDK	API 24+ (Android 7.0); compile/target SDK 37
JDK	17
Disk space	Several GB for downloaded AI models
Device	ARM64 physical device recommended; emulator supported for most features

No NDK, CMake, or native toolchain is required — the SDK ships prebuilt native libraries inside its published AARs.

---

Setup

This sample consumes the RunAnywhere SDK entirely from Maven Central. There is nothing to stage, build, or link locally: a clean clone compiles as soon as Gradle can reach the network.

1. Clone and open the example

git clone https://github.com/RunanywhereAI/runanywhere-android.git
cd runanywhere-android

2. Point Gradle at your Android SDK

Export ANDROID_HOME, or copy local.properties.example to local.properties and set sdk.dir.

3. Verify and run

./scripts/verify.sh

Or open the project in Android Studio and run the app configuration, or install from the command line:

./gradlew :app:installDebug

---

SDK dependency

All SDK artifacts come from Maven Central under the group io.github.sanchitmonga22, pinned by the single runanywhere version in gradle/libs.versions.toml (currently 0.20.18). Nothing is declared as a local AAR or project path:

// gradle/libs.versions.toml
runanywhere = "0.20.18"

// app/build.gradle.kts
implementation(libs.runanywhere.sdk)       // io.github.sanchitmonga22:runanywhere-sdk
implementation(libs.runanywhere.llamacpp)  // io.github.sanchitmonga22:runanywhere-llamacpp
implementation(libs.runanywhere.onnx)      // io.github.sanchitmonga22:runanywhere-onnx
implementation(libs.runanywhere.qhexrt)    // io.github.sanchitmonga22:runanywhere-qhexrt-android

Coordinate	Role	On Maven Central at 0.20.18
runanywhere-sdk	Core SDK + commons native libraries	✅
runanywhere-llamacpp	LlamaCPP backend (LLM, VLM)	✅
runanywhere-onnx	Sherpa-ONNX backend (STT, TTS, VAD)	✅
runanywhere-qhexrt-android	QHexRT backend (Qualcomm Hexagon NPU)	✅

All four are published together; never mix versions. To move to a new SDK release, bump runanywhere in gradle/libs.versions.toml, then regenerate the two reproducibility files that pin the resolved graph — app/gradle.lockfile and gradle/verification-metadata.xml:

# 1. Dependency lock. Host-independent, so any OS will do.
./gradlew :app:dependencies --write-locks

# 2. Checksums. Gradle *merges* into the existing file — it adds entries and never
#    removes them — so run this on top of the committed file rather than deleting it.
GRADLE_USER_HOME="$(mktemp -d)" ./gradlew --write-verification-metadata sha256 \
    :app:assembleDebug :app:testDebugUnitTest :app:lintRelease

Two traps make this file easy to get subtly wrong; both were hit while writing the current one:

• Use a throwaway GRADLE_USER_HOME. Checksums are only recorded for artifacts Gradle actually downloads during the run. Against a warm ~/.gradle the run looks successful but silently omits things already cached — in practice a handful of parent POMs and BOM metadata (guava-parent, junit-bom, kotlin-gradle-plugins-bom). The gap is invisible until someone builds from a genuinely cold cache, i.e. CI.
• Cover Linux and macOS. A few build-time artifacts are OS-classified (com.android.tools.build:aapt2:…-linux.jar vs …-osx.jar) and Gradle records only the host's. The committed file carries both, so one file satisfies the Linux CI runner and a macOS developer. Because step 2 merges, the way to keep both is to run it on Linux (Docker is fine), commit that file, then run it again on macOS on top. If you can only reach one OS, hand-add the missing <artifact> line to the com.android.tools.build:aapt2 component — a Linux-only file breaks every macOS developer, and a macOS-only file breaks CI.

Then confirm the result the same way CI will, with no bypass flags:

./gradlew :app:assembleDebug          # dependency verification live
CI=true ./gradlew :app:assembleDebug  # + LockMode.STRICT

Testing an unreleased SDK build (developer-only)

To try a change from a runanywhere-sdks checkout before it is on Maven Central, publish it to ~/.m2 and point this repo at it:

# In the monorepo — publishes io.github.sanchitmonga22:*:<core/VERSION>
(cd path/to/runanywhere-sdks/bindings/kotlin && ./gradlew publishToMavenLocal)

# Here — opt in per invocation, and relax verification for that one run
./gradlew :app:assembleDebug \
    -Prunanywhere.useLocalSdkAars=true \
    --dependency-verification=lenient

-Prunanywhere.useLocalSdkAars=true adds mavenLocal() ahead of Google and Maven Central, scoped by content { includeGroup("io.github.sanchitmonga22") } so a stale ~/.m2 copy of any other dependency cannot shadow the verified one.

--dependency-verification=lenient is required alongside it, and is not a bypass being smuggled in. A locally published AAR has the same coordinates as the released one but different bytes, so it can never match the sha256 in gradle/verification-metadata.xml. Without the flag the build stops with artifacts failed verification, which is the gate working correctly. Relax it per invocation like this; do not add a <trust group="io.github.sanchitmonga22"/> entry to the committed metadata, because that would permanently un-pin the four artifacts the gate exists to pin.

Both flags are per-invocation only. Never commit runanywhere.useLocalSdkAars to gradle.properties and never set it in CI: ci.yml exists to prove a clean clone resolves the SDK from Maven Central, and a local AAR would make that proof vacuous.

---

Continuous integration

.github/workflows/ci.yml runs on every push to main and every pull request: ubuntu-latest, Temurin JDK 17 (plus JDK 21 for the Gradle daemon, which gradle/gradle-daemon-jvm.properties pins), the Android SDK via android-actions/setup-android (platforms;android-37.0, build-tools;37.0.0), Gradle caching via gradle/actions/setup-gradle, then ./gradlew :app:assembleDebug.

CI runs the unmodified command — no bypass flags — so it exercises exactly the path a developer does. Both reproducibility gates are enforced there and locally:

Gate	What enforces it	What it pins
gradle/verification-metadata.xml	Auto-enabled by Gradle whenever the file exists; ./scripts/verify.sh additionally passes --dependency-verification strict	sha256 of every resolved artifact, including the four io.github.sanchitmonga22 AARs and both OS variants of aapt2
app/gradle.lockfile	app/build.gradle.kts flips to LockMode.STRICT when $CI is set (or with -Prunanywhere.strictLocks=true); LENIENT otherwise so Android Studio sync stays friction-free	the exact resolved version of every module on every configuration

Both files are committed and current. If a dependency or SDK bump makes either gate fail, regenerate them (see SDK dependency) — do not add --dependency-verification=off or env -u CI to the workflow, because that hides the breakage from CI while every clean clone keeps failing.

---

Features

Feature	Description
AI Chat	Streaming LLM conversations with analytics and thinking-mode support
Speech-to-Text	Batch and live transcription via Sherpa-ONNX / Whisper
Text-to-Speech	Neural Piper voices and system TTS fallback
Voice Assistant	Full STT → LLM → TTS pipeline
Vision (VLM)	Camera and image understanding
RAG	Document ingestion and on-device Q&A
Model Management	Download, load, unload, and delete models
Storage	Usage overview and cache cleanup
Solutions	YAML pipeline demos synced from shared catalog
Offline	Inference runs locally after models are downloaded

---

NPU / QHexRT (Snapdragon devices)

On supported Qualcomm Hexagon NPU hardware, the app can register the QHexRT backend for accelerated inference. The QHexRT backend ships as io.github.sanchitmonga22:runanywhere-qhexrt-android and is already on the app's dependency list.

To test private runanywhere/*_HNPU model bundles:

1. Open Settings → Downloads.
2. Enter a Hugging Face token and tap Save token.
3. Download and load an HNPU model from the model picker. The SDK resolves the correct Hexagon architecture natively.
4. Tap Clear to return to public, no-auth downloads.

The token is passed through the SDK at runtime; it is not stored in source, assets, or logs. Private QHexRT release and device-suite workflows live in a separate checkout—see your internal QHexRT documentation if you maintain that stack.

---

Project structure

RunAnywhereAI/
├── app/src/main/java/com/runanywhere/runanywhereai/
│   ├── RunAnywhereApplication.kt    # SDK init and backend registration
│   ├── ui/screens/                  # Feature screens (chat, voice, vision, …)
│   ├── ui/navigation/               # Compose navigation
│   ├── ui/theme/                    # Material 3 theming (#FF6900 brand)
│   └── data/                        # Model catalog, settings repositories
├── gradle/libs.versions.toml         # SDK Maven coordinates + all dependency versions
├── scripts/
│   ├── verify.sh                    # Strict debug APK build gate
│   └── smoke.sh                     # Fast SDK API coverage check
└── README.md

The app resolves every SDK artifact from Maven Central — it contains no local AARs and no relative paths into an SDK source tree, so it builds standalone.

---

Troubleshooting

Symptom	Fix
Could not find io.github.sanchitmonga22:runanywhere-*	Check the runanywhere version in gradle/libs.versions.toml is actually published to Maven Central, and that mavenCentral() is reachable
Gradle dependency verification failures (N artifacts failed verification)	Regenerate gradle/verification-metadata.xml — follow SDK dependency step 2 exactly, including the throwaway GRADLE_USER_HOME and the Linux + macOS passes
... is not part of the dependency lock state (usually only with CI=true)	Regenerate the lock: ./gradlew :app:dependencies --write-locks
QHexRT / NPU models unavailable	Confirm the device has a supported Hexagon NPU; HNPU bundles also require a saved HF token

For a quick static check without a full compile:

./scripts/smoke.sh

---

Related links

Resource	Link
Kotlin SDK	github.com/RunanywhereAI/runanywhere-sdks
Maven Central	io.github.sanchitmonga22
Play Store	com.runanywhere.runanywhereai
Discord	discord.gg/N359FBbDVd
Issues	GitHub Issues
Email	founders@runanywhere.ai

---

License

This project is licensed under the RunAnywhere License (Apache 2.0 based, with additional commercial-use terms). See LICENSE for details.