At the I/O developer conference this week, Google demonstrated how its AI can support more automation in the developer space. Presentations included applying AI to mobile Android development and cloud application deployment.
Studio Bot for Mobile Android Development
Matthew McCullough, VP of Product at Android Developer, has introduced three new features, all powered by AI, to the world of Android development. First, Google added his AI directly into his Android developer workflow. Second, Google has added even more support for developing in a multi-device world, including support for foldable and clamshell phones, which Google also announced at his I/O on Wednesday. Finally, he demonstrated how the new language toolkit and tooling improvements “integrated seamlessly into his modern Android development stack,” McCullough said.
Jamal Eason, Director of Product Management, demos how AI is leveraged in an Android Studio tool called Studio Bot that helps refine code and deploy according to best practices based on the hardware the app is deployed on. Did. Assistant His window located in the toolbar allows developers to chat with and create bots.
“Studio Bot is an AI-powered helper tightly integrated into Android Studio, designed to increase your productivity,” said Eason. “What’s unique about this chat setup is that you don’t have to submit your source code to Google. [it’s] It’s just a chat interaction between you and the bot. ”
Studio Bot can generate code, of course it can, but Eason explained that it can do more than that.
“This is the place to ask contextual questions. For example, this app runs on Jetpack Compose, so I want this layout in Kotlin instead of XML. Let’s ask Studio Bot. “Jetpack Compose How can I do this in ?” And perfect, the code makes sense. ”
It will also provide additional guidance and documentation. Developers can also ask Studio Bot to create unit tests for their app.
“I’m just having a conversation with Studio Bot, and Studio Bot remembers the context between questions,” he said, demonstrating that Studio Bot created unit tests in the right context.
The bot can also explain what caused the crash. In this case Eason forgot to add internet permission. Thanks to AI, the missing code was generated and added with the click of a button.
“Once you’ve built your app with the help of Studio Bot, you’re ready to publish it to Google Play. “So today we are launching a new Google Play Console experiment that regenerates custom store listings for different types of users. Ultimately you control what you send, but Google Play helps you be more creative from start to finish, from development to publishing.We are bringing AI to help our customers act faster and be more creative.”
Cloud development with Duet AI
Cloud development hasn’t been left out of the generative AI rush either. Google introduced Duet AI as an AI-powered collaborator with the ability to build code models trained directly on your own code.
“When it comes to the cloud, generative AI opens the door for professional developers of all skill levels to be more productive,” said Cheng Gold, GM and VP of Engineering, Kubernetes and Serverless at Google. Berg said. “We believe the addition of AI will fundamentally change the way developers of all skill levels build cloud applications.”
With the new cloud capabilities, any developer can build enterprise-ready applications without requiring expertise in security, scalability, sustainability and cost control, he said.
She demonstrated adding Hindi support to a shopping website called Simple, which has many customers in India. She used her Cloud Workstation, a secure, fully managed development environment that is now generally available.
“Just create a function and add a comment. Thanks to Duet AI, I can now see a code snippet for using the Cloud Translation API.” [that] He immediately suggested it to me,” Goldberg said. “Generating code is a good start, but good software engineering practices, like making sure your dependencies are up to date, are essential. You can make sure.”
AI detected that she was running an outdated version of the telemetry library, allowing her to upgrade with a click, and her website now supports Hindi.
“What would have taken me a long time to do, obviously I couldn’t have done it alone, but it was done in a matter of minutes,” Goldberg said.
It can also be applied to existing code.
“One of my personal favorite ways to use Duet AI is to simplify the task of maintaining large code bases,” says Goldberg. “I came across this code and it’s unfamiliar code. Now instead of pinging the owner, searching for relevant code, and spending hours reviewing it, Duet AI will tell you this code. You can help me understand the code part.”
Duet AI is currently only available through Google’s Trusted Tester program. VertexAI offers a similar experience for your own codebase, she added.
“You can use your own codebase to tune and customize Google’s underlying models, no ML expertise required,” she said. “And you can call the custom code model directly from the Duet API.”
Vertex AI can be used to refine and customize the underlying model, but it can also be used to create new content such as images, she said.
“Vertex AI gives you easy access to the full suite of underlying models from Google and our open source partners, with enterprise-grade data governance and security, without worrying about all the work required to set them up.” Goldberg said. demonstrated the ability of his Vertex AI to take a picture of a handbag and add it to an image-based model to create multiple variations of the image.
“It works regardless of image complexity, giving you the freedom to easily iterate and explore different options without the complexity of hosting your own model or understanding hyperparameters. she said. “With Vertex AI, we can upgrade quickly and easily.” [the image] So it looks consistent on high resolution displays in your online store as well as in print, and you’re almost ready to add it to your site. As I expand my business globally, I will be harnessing the power of Vertex AI to generate text captions for accessibility and localize them into over 300 languages. ”
Look, mama, no code
She then went to great lengths to show how Duet AI can be integrated with Google Workspace to create apps without even knowing how to code.
“Describe in natural language the travel authorization app you want to build. Then Duet AI walks you through the process and asks you simple questions like ‘How do you want to be notified?’ What are the main sections of the app? And most importantly, what is the name of the app? Let’s call it Simple Travel,” she said. “Answer the question and Duet AI will create an app with travel requests from my team within Google Workspace.”
He added that the new chat API is now available in Google Workspace and will be generally available in the coming weeks. These APIs allow developers to build chat apps that allow users to perform actions such as creating and updating records. She pointed out that Atlassian used these APIs to build her Jira app for chat. Jira apps help teams track issues, manage projects, and automate workflows.
He added that a new Google Meet API and two new SDKs will be previewed in the coming weeks to allow developers to bring Google Meet data and functionality into their apps.
For the AI models underlying these developments, see Google’s new TensorFlow tools and approach to ML fine-tuning.
