Follow ZDNET: Add us as a preferred source on Google.

ZDNET’s key takeaways

• Google is expanding AI across the Android Studio workflow.
• Google intends AI to reduce “toil” for work like tests, updates, and API migrations.
• Google says coding shifts from “how” to defining “what.”

While it may seem that Anthropic’s Claude Code has been capturing all of the AI coding attention, enormous development strides are also being made by the other big AI players.

Google, for example, has been working hard to expand AI-based development capabilities for Android developers. ZDNET recently had the opportunity to talk with Sam Bright, VP and GM of Google Play and Developer Ecosystem at Google.

Also: 5 custom ChatGPT instructions I use to get better AI results – faster

We discussed the use of Google’s Gemini generative AI technology and how it’s being incorporated into Android Studio. This is providing AI capabilities across the entire Android development workflow.

Interestingly, Google isn’t locking developers into just Gemini, although it’s certainly proud of its increasingly impressive LLM offerings. Google is allowing developers to choose which large language model powers the AI features inside the IDE. This is similar to the way Apple ecosystem developers can choose their preferred model inside Xcode.

Bright also discussed adding enterprise-level features like enterprise-grade privacy, security, and repository-aware customization to Gemini’s business tier.

The big thought: AI shifts developers from ‘how’ to ‘what’

Bright told ZDNET, “Looking ahead three to five years, the day-to-day work of an Android developer will shift from writing ‘how’ to describing ‘what.'” He contends that the daily work of coders will move from writing implementation details to defining intent and outcomes.

Also: Claude Sonnet 4.6 delivers frontier-level AI for free and cheap-seat users

In my experience with using AI to build apps, that’s certainly true… to a point. While my available programmer time hasn’t been spent writing code line by line, I have spent a substantial amount of my time not just specifying what I want the AI to do, but figuring out what the AI missed.

For example, I have an app that is supposed to sync between mobile and desktop implementations. It does, but not completely. The AI didn’t point out where it missed sync operations. That was something I had to use my programming and problem-solving skills to discover, narrow down, and then present to the AI for bug fixing.

That said, Google is positioning “AI as a partner that helps them move faster.” That is something I have definitely experienced. Work that would have taken me years without AI has been completed in days or weeks.

Reducing toil through agentic automation

Bright is very big on the word “toil” to describe the work AI can help with. Specifically, he mentioned work like:

• Necessary but tedious steps that most developers would prefer to skip.
• Repetitive tasks such as writing boilerplate tests or updating dependencies.
• The manual work of “pixel-pushing.”
• High-friction, low-creativity tasks, which include upgrading dependencies, migrating deprecated APIs, or writing standard unit tests.
• Tasks across the entire application lifecycle, such as keeping libraries updated.
• Mechanical tasks that AI can handle on behalf of developers.

I asked him, “How do you decide which tasks should be automated versus those that should remain firmly in the developer’s hands?”

This is important because some work, particularly the vision and direction of the project, needs to stay in human hands. Also, as we’ve seen, overseeing the work of the AI and finding blind spots also needs to be something that’s in the human programmer’s hands.

Continuing in the theme of toil, Bright told me, “When we’re deciding what to automate, we look for toil. Those are tasks that usually kill a developer’s momentum and don’t require a creative spark.”

Also: I got 4 years of product development done in 4 days for $200, and I’m still stunned

Those tasks are also the ones that can take a ton of time and can cause you to question why you went into programming to begin with. I spent a couple of years using all my available programming time to do pretty much nothing else except update APIs because the platform vendor kept changing them. My profanity skills increased exponentially.

Bright told me that his team is also concerned about what it is the AI is doing when set loose into a code base. “We also prioritize transparency so AI isn’t making changes on its own, but suggesting improvements that fit right into the code reviews and workflows a developer already uses.” He explained, “This keeps them in the driver’s seat, making decisions on the direction of the project while AI just handles the manual heavy lifting.”

Enterprise-grade flexibility, privacy, and control

Bright said, “Every developer has a unique workflow when using AI, and different companies have varying policies on AI model usage.” That’s why it’s enabled the ability for developers to choose which LLM they wish to use in Android Studio. That gives them control over performance, privacy, and cost. Context-aware agents generate project-specific suggestions that compile and respect architecture and style.

The new business tier is “backed by Google Cloud.” It is offered with security features like Private Google Access, VPC Service Controls, and Enterprise Access Controls with granular identity and access management permissions to help enterprises adopt AI assistance at scale without compromising on security and privacy. Bright recommended looking at Google’s data governance policy if you’re curious about privacy issues.

Also: The AI coding gap: Why senior devs are getting faster while juniors spin their wheels

Allaying a concern about code usage and privacy use by Gemini, Bright told ZDNET, “Our policies help ensure customer code, customers’ inputs, as well as the recommendations generated will not be used to train any shared models.”

“Teams also need to trust that their proprietary code stays theirs,” Bright continued. “That’s why no code is stored, and crucially, a business’s code is never used for model training.”

From rapid prototyping to long-term maintenance

AI systems can dramatically compress design-to-prototype timelines. Bright says, “Instead of a developer spending days translating a Figma design into code just to see if it feels right, they can now get to a functional prototype in minutes. It drastically tightens the feedback loop.”

He cited Entri, an online learning app, as an example. “What typically took them hours of UI work was completed in just 45 minutes. This allowed them to cut their average UI build time by 40%.”

Also: 6 essential rules for unleashing AI on your software development process – and the No. 1 risk

But a developer’s job doesn’t end after cranking out a new app or new piece of code. Far from it. In fact, much of a developer’s work effort is spent maintaining and updating that code. Google’s Version Upgrade Agent is designed to help a developer update dependencies. It does this by analyzing their project, checking for potential updates, and then attempting to apply them.

Bright told ZDNET, “This can save developers a significant amount of time and effort. And for an application developer, every investment you make to stay up to date is a strategic advantage.”

The AI in Android Studio can also help with maintenance by helping to analyze crash reports and suggesting targeted fixes.

Maximizing innovation while reducing toil

Bright told ZDNET, “On the professional Android developer side, we expect to see an increase in app quality as AI handles a lot of the mechanical toil on behalf of developers.”

He told us AI will help developers follow modern best practices without having to manually track every minor update. AI can help identify and resolve performance bottlenecks or crashes before they ever reach a user, especially as the AI systems gain better integration with developer tools like Google’s App Quality Insights.

AI can help developers do more testing, again because the toil in doing so is substantially reduced. The result is that apps are less likely to launch with bugs and will feel a lot more stable across a range of devices.

Bright also sees vibe coding opening doors. “We’re seeing more people start developing apps, and it’s great to see that barrier of entry lowered,” he told ZDNET. “This shift means the next generation of Android apps will come from a much more diverse group of creators.”

Also: The best free AI courses and certificates for upskilling in 2026 – and I’ve tried them all

The big picture, said Bright, is that “When the basics are backed by AI, it frees up developers to focus on actual innovation.”

He said, “They can spend their energy on the unique features and creative experiences that make an app special, rather than getting stuck in the weeds. It’s really about giving the apps themselves a better foundation so they can be more capable and reliable for everyone sooner rather than later.”

I know that AI-based coding has helped me create apps in timeframes that would have been absolutely impossible otherwise. What about you?

Are you using AI inside Android Studio or another IDE as part of your daily workflow? Has AI reduced the “toil” in your projects, or do you find yourself spending just as much time reviewing and correcting what it produces? Do you see the shift from writing “how” to defining “what” happening in your own development work? And how important are model choice, privacy guarantees, and enterprise controls in your decision to adopt these tools? Let us know in the comments below.

You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.