Building a Mobile Crypto News Aggregator App

Building a Mobile Crypto News Aggregator

Crypto news aggregation means juggling multiple real-time data streams: RSS feeds, APIs (CryptoCompare News, CoinGecko, Messari), Twitter/X, and price data. The challenge: collect everything, deduplicate, rank by relevance, and deliver to the user without delay. We've been tackling this since the demand for aggregators first emerged, building reliable mobile solutions for FinTech. Our experience: 5+ years and 15+ launched apps. Significant savings on server resources are achieved through caching and deduplication. Contact us for a consultation to discuss your project.

Data Source Comparison and Selection

CryptoCompare News API — the free tier gives 100k requests per month, supports filtering by categories (Bitcoin, Ethereum, DeFi, NFT) and languages. Messari API offers more analytical content, good for serious financial news, but the full access requires a paid tier. RSS feeds — CoinDesk, Decrypt, The Block — publish RSS; we parse server-side using feedparser (Python) or custom RSS parsers.

How to Avoid News Duplication?

Duplication occurs when the same news arrives from different sources (e.g., CryptoCompare and Messari). We apply server-side deduplication: compare headlines using difflib (Python) or Levenshtein distance. A similarity threshold of 0.85 catches 95% of duplicates. This cleans the feed and reduces client requests. In a recent project for a major crypto analytics platform, we reduced duplicate noise from 80% to under 5% and improved feed delivery speed from 8 seconds to 1.2 seconds by combining this deduplication with a Redis-backed cache.

How to Achieve Real-Time Updates?

News appears constantly. Polling every 60 seconds is a reasonable compromise for most users. WebSocket for real-time updates — if you need sub-second reaction (e.g., breaking news).

WebSocket is 10x faster than polling for real-time, but requires more server resources. The choice depends on your budget and freshness requirements. If you need a real-time solution, contact us — we'll design the optimal architecture.

class NewsWebSocketClient: ObservableObject {
    @Published var latestNews: [NewsItem] = []
    private var webSocketTask: URLSessionWebSocketTask?

    func connect() {
        let url = URL(string: "wss://api.yourservice.com/news/stream")!
        var request = URLRequest(url: url)
        request.addValue("Bearer \(token)", forHTTPHeaderField: "Authorization")
        webSocketTask = URLSession.shared.webSocketTask(with: request)
        webSocketTask?.resume()
        listen()
    }

    private func listen() {
        webSocketTask?.receive { [weak self] result in
            if case .success(.string(let text)) = result,
               let item = try? JSONDecoder().decode(NewsItem.self, from: Data(text.utf8)) {
                DispatchQueue.main.async {
                    self?.latestNews.insert(item, at: 0)
                    if self?.latestNews.count ?? 0 > 200 {
                        self?.latestNews.removeLast()
                    }
                }
            }
            self?.listen()
        }
    }
}

Personalization and Filtering

Users follow specific coins — BTC, ETH, SOL. Client-side filtering by tags works only if all news is already loaded. For large volumes, we use server-side filtering with query parameters. We guarantee the feed loads in 1–2 seconds even with 10,000 cached news items. API cost optimization — we choose the right tier based on traffic volume.

AI Summarization: for each news article, we generate a 2–3 sentence summary via GPT-4o mini on the server during indexing. This is more costly than showing the original lead, but users get the essence without opening the article. The summarization cost is economical — we select the optimal model for your budget.

Headline sentiment analysis (Positive / Neutral / Negative) provides a quick market signal. We use ML Kit Sentiment on Android or NLTagger.sentimentScore on iOS, or specialized crypto sentiment models (CryptoNewsScore).

How to Show Quotes Alongside News?

Displaying the current price of a coin next to a news item is standard for crypto aggregators. CoinGecko free API can fetch prices for 250+ coins with a single request /simple/price?ids=bitcoin,ethereum&vs_currencies=usd,rub. A 30-second cache prevents over-fetching.

// Android: correlate news with price widget
data class NewsWithPrice(
    val news: NewsItem,
    val relatedCoin: CoinPrice?  // null if coin not identified
)

fun enrichNewsWithPrices(
    news: List<NewsItem>,
    prices: Map<String, CoinPrice>
): List<NewsWithPrice> {
    return news.map { item ->
        val coin = item.tags.firstOrNull { prices.containsKey(it.lowercase()) }
        NewsWithPrice(news = item, relatedCoin = coin?.let { prices[it.lowercase()] })
    }
}

Push Notifications for Breaking News

Breaking news alerts — by keywords or coins from the user's watchlist. Firebase Cloud Messaging with topic subscriptions: /topics/bitcoin-news, /topics/ethereum-news. Users subscribe in the app settings. We recommend no more than 3–5 pushes per day to avoid unsubscribes. Background fetch (iOS) or WorkManager (Android) for silent feed updates. We configure notifications to be useful, not intrusive.

Our Work Process

1. Analysis — agree on sources, personalization requirements, data volume.
2. Design — server architecture, APIs, data schemas.
3. Implementation — server aggregation, deduplication, AI summarization, mobile client.
4. Testing — load testing with 1000+ news items per day, latency checks.
5. Deployment — publish to App Store and Google Play, set up monitoring.

What's Included in the Work (Deliverables)

• Server API documentation.
• Access to Firebase, CoinGecko, selected sources.
• Source code under NDA.
• Training of client team on the admin panel.
• 2 weeks of post-launch support.

Timeline Estimates

Basic aggregator with multiple sources and filters — 4–6 weeks. Full system with AI summarization, sentiment, personalization, and push notifications — 10–16 weeks. Timelines vary based on requirements.

We'll evaluate your project for free — contact us for a consultation. Turnkey development ensures transparency and on-time delivery. Discuss your project with our engineer — we'll find the optimal solution.

Machine Learning in Mobile Apps: CoreML, TFLite, and On-Device Models

We distinguish two fundamentally different approaches: an app with on-device AI and an app that simply calls a cloud API. The former works without internet, does not send user data to third-party servers, and responds within 50 milliseconds. The latter depends on network latency and pricing plans. Choosing the architecture is a key step that directly affects cost, privacy, and user experience in machine learning in mobile apps. Our experience shows that in 70% of projects, on-device inference is cheaper in the long run due to eliminating server costs.

How to Choose Between CoreML and TFLite for On-Device Inference?

CoreML — Apple's native framework for running ML models on device. Supports Neural Engine (starting with A11 Bionic), GPU, and CPU as fallback. Models are converted to .mlmodel format via coremltools from PyTorch, ONNX, or TensorFlow. Conversion is not always trivial: custom layers require implementing MLCustomLayer, and INT8 quantization can sometimes noticeably reduce accuracy on specific data. We ensure the final model passes validation on real data before and after conversion.

TensorFlow Lite — cross-platform alternative for Android and Flutter. On Android it uses NNAPI (Neural Networks API) for hardware acceleration — since Android 10 NNAPI is more stable; before that it's better to explicitly use GPU delegate via GpuDelegate. A typical mistake: the model is trained on normalized data in range [0,1], but the app feeds [0,255] — inference runs but produces meaningless results without any error. We include an automatic input data validation module in the SDK.

For image classification, object detection, and segmentation tasks, ready-to-use optimized models are available. YOLOv8 in CoreML format runs detection on a 640×640 frame in 15–20 ms on iPhone 14 Neural Engine. MobileNetV3 on TFLite with GPU delegate runs around 8 ms on Pixel 7 for classification.

What If You Need Text Generation On-Device?

Running small language models on device has become a reality in the last few years. Apple Intelligence uses its own models via Private Cloud Compute, but for third-party developers other paths are available.

llama.cpp with Metal backend on iOS is a working approach for phi-3-mini (3.8B parameters, 4-bit quantization, ~2.3 GB). Inference: 15–25 tokens/second on iPhone 15 Pro. For integration in Swift, use the Swift Package llama.swift or a wrapper via C interface llama.h. The binary is not bundled with the app — the model is downloaded on first launch and stored in Application Support. Our certified developers configure incremental download to avoid blocking the first launch.

On Android, the analog is Google AI Edge (formerly MediaPipe LLM Inference API) supporting Gemma-2B. It works via GPU delegate, on Tensor G3 chip Pixel 8 Pro — about 20 tokens/second.

Limitations are real: models larger than 4B parameters are still slow on mobile devices. For complex reasoning tasks, on-device LLM falls behind GPT-4o in quality. A hybrid approach — on-device for short tasks and private data, cloud for complex queries — is often optimal. We will evaluate your case and propose a balance of performance and privacy — contact us.

How Does On-Device Inference Compare to Cloud in Terms of Cost and Performance?

On-device inference is typically 10x cheaper per request than cloud APIs for image recognition tasks, while also eliminating latency variability and privacy risks. The table below summarizes the trade-offs.

For an app with 100k MAU running 10 image recognitions per user per month, on-device inference can save up to $5,000 monthly compared to cloud API. Get a free consultation on your ML architecture today.

Integrating OpenAI API and Other Cloud Models

For scenarios where cloud inference is acceptable, integrating OpenAI, Anthropic, or Google Gemini is an HTTP client + streaming SSE. In Swift, AsyncThrowingStream is convenient for streaming responses. In Kotlin, use Flow.

Critically: API keys must never be stored in the app bundle. Even an obfuscated key can be extracted from the IPA in 10 minutes using strings or frida. Correct architecture: mobile app → your own backend → OpenAI API. The backend controls rate limiting, logs requests, and protects the key.

What Is Included in the Work (Deliverables)

• Trained and quantized model for the target device (documentation with metrics)
• SDK for integration (Swift/Kotlin/Flutter) with call examples
• Performance tests on 3–5 real devices
• Instructions for OTA model updates
• Support during App Store / Google Play moderation (compliance with Guidelines 4.2, 5.1)
• 2 weeks of technical support after release

Typical Project Pipeline

1. Task analysis — measure latency, privacy, size, supported devices.
2. Model prototyping — in Python, evaluate accuracy on target data.
3. Conversion and quantization — for CoreML/TFLite with validation.
4. Integration into the app — model wrapped in a service layer (easy to swap CoreML ↔ TFLite ↔ cloud).
5. Testing — on real devices, measure FPS, RAM, battery.
6. Deployment — via TestFlight / Firebase App Distribution, monitor metrics.

Timelines: integration of a ready CoreML/TFLite model — 1–2 weeks, development of a custom model with mobile optimization — from 6 weeks, on-device LLM chat with personalization — 4–8 weeks.

Why We Take on Complex Cases?

10+ years of experience in mobile development, 50+ implemented AI/ML solutions, guarantee of compatibility with current iOS and Android versions. All projects undergo code review and load testing. The cost includes preparation of moderation documentation and training of your team.

Contact us — we will help you choose the architecture and implement ML in your app turnkey. Order an audit of your existing solution — we will assess the potential for server cost savings free of charge. In some projects, savings can reach significant amounts per month.