This is Part 1 of a series of articles where I explain how to implement GenAI on Android. [Click here to view the full series.]

Making use of AI to summarise text

I thought on-device AI was still a distant dream — until I plugged ML Kit’s new GenAI summarisation API into my side-project, SmartWriter. Five minutes later, my phone was generating smart summaries from messy blocks of text — with zero cloud calls and no OpenAI keys.

In this post, I’ll show you exactly how I built it using Jetpack Compose and MVVM — and you can try it yourself because the full code is live here:

👉 GitHub: SmartWriter

📲 What we’re building

The app has a Summarise screen:

Paste or type a long message, hit a button, and get a smart, clean summary — all processed locally on your phone.

🔍 Example:

Input:

Hey John, just a quick reminder that we’ve got the product demo with the client at 10:00 a.m. tomorrow. Make sure you’ve updated the pitch deck and double-checked the data from last week. Also, the UX team had a few notes about the prototype that might be worth bringing up.

Output:

Reminder: Client demo at 10 a.m. tomorrow. Update pitch deck and check UX notes.

🧱 Project setup

Add this dependency to your libs.versions.toml:

mlkit-genai-summarization = "com.google.mlkit:genai-summarization:1.0.0-beta1"

Then in your build.gradle.kts:

dependencies {
    implementation(libs.mlkit.genai.summarization)
}

✅ You must use Android 14+

📱 Only works on compatible devices (e.g. Galaxy S25 Ultra, Pixel 8 Pro) that have hardware support for AI. Emulators don’t support Gemini Nano (yet).

🧩 The UI (Jetpack Compose)

Here’s a simplified version of my SummarisationScreen:

@Composable
fun SummarisationScreen(
    uiState: SummarisationUiState,
    onInputTextChanged: (String) -> Unit,
    onSummariseClicked: (Context) -> Unit,
) {
    val context = LocalContext.current
    
    Column(
        modifier = Modifier
            .fillMaxSize()
            .padding(16.dp)
            .verticalScroll(rememberScrollState())
    ) {
        OutlinedTextField(
            value = uiState.inputText,
            onValueChange = onInputTextChanged,
            label = { Text("Enter text to summarise") },
            modifier = Modifier.fillMaxWidth()
        )
        Spacer(modifier = Modifier.height(16.dp))
        Button(
            onClick = { onSummariseClicked(context) },
            enabled = uiState.inputText.isNotBlank() && !uiState.isLoading
        ) {
            if (uiState.isLoading) CircularProgressIndicator(color = Color.White)
            else Text("Summarise")
        }
        Spacer(modifier = Modifier.height(24.dp))
        if (uiState.summary.isNotBlank()) {
            Text("Summary:", style = MaterialTheme.typography.titleMedium)
            Spacer(modifier = Modifier.height(8.dp))
            Text(uiState.summary)
        }
    }
}

🧠 ViewModel

This version includes a simplified version of the viewmodel logic. You can find the full implementation in the GitHub repo.

fun onSummariseClicked(context: Context) = viewModelScope.launch {
    // I chose these options but feel free to inject these or use others
    val options = SummarizerOptions.builder(context)
        .setInputType(InputType.ARTICLE)
        .setOutputType(OutputType.ONE_BULLET)
        .setLanguage(Language.ENGLISH)
        .build()    

    val summarizer = Summarization.getClient(options)
    val status = summarizer.checkFeatureStatus().await()
    
    when (status) {
        FeatureStatus.AVAILABLE -> {
            val request = SummarizationRequest.builder(uiState.value.inputText).build()
            _uiState.update { it.copy(isLoading = true) }
           val summary = summarizer.runInference(request).await().summary
            _uiState.update { it.copy(summary = summary, isLoading = false) }
        }

        FeatureStatus.DOWNLOADABLE -> {
            // visit the GitHub repository for the full implementation
            summarizer.downloadFeature().await()
            onSummariseClicked(context) // retry after download
        }
        
        else -> {
            _uiEvent.emit(SummarisationUiEvent.Error("Your device does not support this feature."))
        }
    }
}

This article was previously published on proandroiddev.com