Introduction to AI in Flutter

Overview

Artificial Intelligence is transforming mobile apps — from face filters and voice assistants to smart recommendation engines. Flutter, being a cross-platform framework, provides excellent support for integrating AI/ML capabilities into apps that run on Android, iOS, web, and desktop.

With Flutter, you can:

• Run on-device ML models for real-time inference (no internet required)
• Use Google ML Kit for ready-made vision and language models
• Load custom TensorFlow Lite models for specialized tasks
• Connect to cloud AI APIs (OpenAI, Google Cloud AI, AWS SageMaker)
• Build Generative AI features with the Gemini API

Why AI in Flutter?

Flutter's architecture makes it uniquely suited for AI-powered apps:

ML Options for Flutter

There are several approaches to adding AI to your Flutter app:

Google ML Kit

Google ML Kit provides ready-to-use machine learning APIs that work offline on-device. It's the easiest way to add AI to Flutter apps.

Supported Features

dependencies:
  flutter:
    sdk: flutter
  google_mlkit_text_recognition: ^0.13.0
  google_mlkit_barcode_scanning: ^0.13.0

TensorFlow Lite

TensorFlow Lite (TFLite) lets you run custom ML models in Flutter. Use it when ML Kit's pre-built models don't cover your use case.

How TFLite Works in Flutter

1. Train a model in TensorFlow (Python) or download a pre-trained model
2. Convert the model to TFLite format (.tflite file)
3. Bundle the model with your Flutter app
4. Run inference using the tflite_flutter package

dependencies:
  flutter:
    sdk: flutter
  tflite_flutter: ^0.10.4

import 'package:tflite_flutter/tflite_flutter.dart';

class ImageClassifier {
  late Interpreter _interpreter;

  Future<void> loadModel() async {
    _interpreter = await Interpreter.fromAsset('assets/model/mobilenet_v1.tflite');
  }

  List<double> classify(List<double> input) {
    var output = List<double>.filled(1001, 0.0).reshape([1, 1001]);
    _interpreter.run([input], output);
    return output[0];
  }

  void dispose() => _interpreter.close();
}

On-Device vs Cloud AI

Choosing between on-device and cloud-based AI depends on your app's requirements:

Real-World Example: Smart Notes App

Imagine building a "Smart Notes" app that uses AI to:

• 📸 OCR: Extract text from photos of handwritten notes
• 🏷️ Auto-tagging: Label notes by topic using image classification
• 🔍 Smart search: Find notes by describing their content
• 📝 Summarization: Use Gemini API to summarize long notes

import 'package:google_mlkit_text_recognition/google_mlkit_text_recognition.dart';

class SmartNotesService {
  final _textRecognizer = TextRecognizer();

  /// Extract text from an image (OCR)
  Future<String> extractTextFromImage(String imagePath) async {
    final inputImage = InputImage.fromFilePath(imagePath);
    final recognizedText = await _textRecognizer.processImage(inputImage);
    return recognizedText.text;
  }

  /// Auto-generate tags from text content
  List<String> generateTags(String text) {
    final keywords = <String>[];
    if (text.contains('function') || text.contains('class')) {
      keywords.add('Programming');
    }
    if (text.contains('equation') || text.contains('formula')) {
      keywords.add('Mathematics');
    }
    return keywords;
  }

  void dispose() => _textRecognizer.close();
}

Summary

• Flutter supports AI through ML Kit, TFLite, and cloud APIs
• Google ML Kit provides free, on-device ML for common tasks
• TensorFlow Lite enables custom model inference
• On-device AI offers low latency and privacy; cloud AI offers more power
• A hybrid approach (on-device + cloud) is often the best strategy