Mobile 2023: The Year AI Came to the App

It's December 2023, and I'm looking back at the year. In January, we were still debating whether AI could actually work in mobile apps. By December, AI features are in the apps most people use every day. That shift happened quietly, but it's significant.

Siri got smarter—not revolutionary, but noticeably better at understanding what you're asking. Google Bard integrated with Gmail, Docs, and Sheets. Apple released Apple Intelligence features that work on-device. And the ChatGPT app became a top-10 app on iOS within weeks of launch.

This wasn't inevitable. Mobile apps are different from desktop applications. Lower computational resources, less screen space, slower networks, intermittent connectivity. Adding AI to mobile required solving those problems. By the end of 2023, the solutions exist. And adoption is real.

What Actually Works

The AI features that gained traction in 2023:

Voice interaction. People use Siri more than they did in 2022. Not because Siri got perfect—it didn't—but because it got incrementally better and people found it genuinely useful for specific tasks. Setting reminders, making calls, sending messages. These work better with voice than typing.

Smart replies. Gmail's smart reply feature was released years ago, but it improved this year. Your phone suggests three responses to emails. For many emails, one of those suggestions is exactly what you want. You tap it, done. That's just convenient enough that people use it.

On-device processing. Apple's move to on-device AI (announced this year) is significant. Your phone processes AI locally. Data never leaves your device. It's faster and more private. Users get better performance and don't worry about privacy.

Search assistance. Google's Bard integration into search is helpful. You ask a question, you get an AI-generated summary along with links. It's faster than scrolling through results.

Writing assistance. Grammarly and similar tools got better at real-time correction and improvement. Not perfect, but good enough that people trust it.

These aren't dramatic. But they're all real. People use them. They save time.

What Didn't Work

Chat interfaces. For general-purpose tasks, chat interfaces on mobile are awkward. Users don't want to hold a conversation with their phone. They want answers. The ChatGPT app is popular, but mostly for specific use cases—technical questions, homework help, brainstorming. Not for general work.

Image generation in apps. Several apps added image generation this year. Adoption was low. The latency is too high, the screen space isn't great, and generating images on a phone isn't how most people work.

Autonomous features. "Let AI do this automatically" rarely works. Users want to control what happens in their apps. They don't trust AI to act autonomously. And on mobile, the stakes feel higher—your phone is personal.

Real-time collaborative AI. The dream of "AI helps your team collaborate in real-time" didn't materialize. The technology exists, but adoption is low. People still work in silos more often than together.

What This Means for Mobile Development

As a builder, here's what 2023 taught me:

AI on mobile is most useful when it's invisible. Users shouldn't have to think about whether something is AI-powered. It just works. The app understands what they're trying to do and helps.

Privacy matters more on mobile. People are more paranoid about their phone than their laptop. On-device processing, local data, no cloud sync—these aren't nice-to-haves. They're expectations.

Speed is non-negotiable. If an AI feature takes 3+ seconds, users abandon it. Mobile users expect instant responses. That means either local processing or very fast inference.

Task-specific AI beats general-purpose AI. A narrow AI that does one thing well is more useful than a general assistant that does many things okay.

For Enterprise Mobile

For ERP and business apps on mobile, 2023 was quieter than 2022. Enterprise didn't rush to add AI. But the experiments we did were successful:

• Voice commands for data entry (faster than typing)
• Computer vision for barcode-less item identification (works well)
• Smart notification aggregation (reduces alert fatigue)
• Anomaly flagging on inventory or orders (catches problems)

These aren't flashy, but they're useful. And they're deployed in production, working daily.

For Arvension, 2023 validated our approach: AI should make mobile work faster, not add new capabilities. A warehouse worker who can speak an order instead of typing it, that's valuable. A finance team member who can point their phone at an invoice and have it auto-categorized, that's valuable. A manager who gets alerted only to orders that actually need their attention, that's valuable.

The Broader Pattern

Looking at 2023 as a whole: AI in mobile shifted from hype to utility. The features that worked were the quiet ones. The ones that save time on repetitive tasks. The ones that users don't even think about.

This is a good sign. When AI becomes boring—when it's just part of how apps work instead of a feature to brag about—that's when it's actually working.

What 2024 Might Look Like

If the pattern holds, 2024 will see more of this. More AI that's invisible. More task-specific assistants. More on-device processing. More integration with enterprise systems.

I'm watching:

• Local AI models improving (Llama, Mistral, open-source tools)
• Better frameworks for on-device ML (Core ML, Dart ML, TensorFlow Lite)
• More enterprise apps with AI built-in, not bolted-on
• Voice interfaces improving (finally)
• Workplace productivity apps leveraging AI more effectively

The exciting stuff isn't what gets headlines. It's the incremental improvements that save hours per month across millions of users.

The Year-End Assessment

2023 was the year AI matured on mobile. Not the year it transformed mobile—that's still coming. But the year it became practical, useful, and boring. The year you could build an AI feature on mobile and know it would work, not just hope it would.

For builders, that's good news. It means we can stop asking "is AI possible on mobile?" and start asking "how do we make AI useful for this specific problem?" Those are better questions.

Looking at the year, I'm optimistic. Not because AI solved everything—it didn't. But because it solved some things well. And by the end of 2023, the trajectory is clear. AI is going to be fundamental to mobile development going forward. Not magic. Not transformation. Just fundamentally better tools for building applications that help people work.

That's a win.