chromeunboxed.com

Google Handed App Migration to Agents — Three Declarations from I/O 2026

Google I/O 2026, Migration Agent, Managed Agents, WebMCP, Android Studio, Gemini API, KotlinDev

17 Things to Know for Android Developers at Google I/O 2026

All the news from the Google I/O 2026 Developer keynote

Managed Agents Quickstart — Gemini API

Moving an iOS app to Android typically takes a team 3–4 weeks. React Native might be faster, but you're still mapping UI components, converting assets, and rewriting in Jetpack Compose by hand. Google declared that its new Migration Agent cuts that down to hours.

Three announcements came out of the Google I/O 2026 Developer Keynote at the same time: Migration Agent, Managed Agents, and WebMCP. Individually, they look like quality-of-life upgrades. Together, they tell one story — AI is moving from assisting development to actually executing it.

3-second summary

iOS / RN code → Migration Agent → Kotlin app + Managed Agents API + WebMCP standard

What are these three things, exactly?

Here's the short version of each announcement.

Migration Agent (Android Studio)
An AI agent that converts iOS, React Native, and web framework code into native Kotlin Android apps. It intelligently maps features, converts assets like storyboards and SVGs, and applies Jetpack Compose best practices. Google claims it turns weeks of manual porting into hours.
Managed Agents (Gemini API)
Get a fully provisioned AI agent from a single API call. A Linux sandbox spins up automatically, pre-equipped with web browsing, code execution, and file management. Built for developers who want to deploy agents without setting up infrastructure. Compute is free during the preview period.
WebMCP (Chrome 149 Origin Trial)
A proposed web standard that changes how browser-based AI agents interact with websites. Instead of simulating clicks, WebMCP lets websites declare structured tools — agents call them directly with speed and reliability. Chrome 149 origin trial starts now.

The common thread across all three: agents are executing development work, not just helping with it.

Hours

Migration Agent time (was weeks)

1 API call

Full agent provisioning

Chrome 149

WebMCP origin trial start

What actually changes?

	Before	After I/O 2026
App migration	Manual rewrite by developer (weeks)	Migration Agent auto-converts (hours)
AI agent deployment	Build server, sandbox, integrations yourself	Single API call, fully provisioned
Web-agent interaction	Click simulation (brittle, slow)	Structured WebMCP tool calls (precise)
New Android apps	Set up project in Android Studio manually	Prompt in Google AI Studio → Kotlin app

Migration Agent isn't just translating code. It analyzes intent, maps features intelligently, and reimplements using Android architecture. It's closer to redesign than translation. Complex business logic will still need developer review, but the structural heavy lifting is done.

With Managed Agents, the key is infrastructure abstraction. Standing up an AI agent used to mean cloud instances, sandbox configuration, tool integration, auth management. Now it's an API key and one line of code. The underlying model is antigravity-preview-05-2026, with reasoning, code execution, and web browsing built in.

WebMCP is the longer-term bet. Think about how often AI agents fail at booking flights or completing checkout flows right now. WebMCP fixes that by letting sites like Expedia declare a search_flights tool — agents call it directly, no click simulation needed. Expedia, Booking.com, Shopify, and Instacart are already testing it.

Managed Agents pricing

During the preview, Linux sandbox compute (CPU, memory, execution environment) is free. You only pay for token usage and tool calls at standard Gemini 3.5 Flash rates. Great time to experiment.

How to get started

Migration Agent — Update Android Studio
Update to the latest Android Studio, then go to File → Migrate → [Source Platform] to Android. Open your existing iOS, React Native, or web project and run the agent.
Managed Agents — Get a Gemini API key and call one function
Grab a Gemini API key and call interactions.create() in the Python or JavaScript SDK. Model: antigravity-preview-05-2026. Compute is free during preview.
WebMCP — Register for origin trial and declare tools
Register for the Chrome 149 origin trial. Choose the Declarative API (annotations on HTML forms) or Imperative API (JavaScript tool registration). Google's travel booking and pizza order demos are on GitHub.
Test WebMCP locally first
Even before Chrome 149, enable chrome://flags/#enable-webmcp-testing to experiment locally.
Build an Android app from Google AI Studio
Go to ai.google.dev, prompt your app idea, and deploy via the Antigravity CLI. Output is Kotlin + Jetpack Compose.

FAQ

Can Migration Agent handle complex business logic automatically?

Currently, it covers feature mapping, asset conversion, and Jetpack Compose best practices. Complex business logic and custom API integrations will likely still need developer review. The 'hours' figure is realistic for simple to medium complexity apps.

Is Anthropic's MCP the same as Google's WebMCP?

They're completely different standards. Anthropic's MCP (Model Context Protocol) is a general-purpose protocol for AI clients to call external server tools. WebMCP is a browser-specific standard for AI agents to use structured tools declared by websites. The similar names are confusing, but they operate at different layers.

What's the best use case for Managed Agents?

It's ideal for quickly experimenting with agents or building prototypes without infrastructure overhead. Since Google manages the execution environment, teams who need custom infra or full execution-layer control should use the Antigravity SDK directly.

What happens to existing websites when WebMCP becomes a standard?

Sites without WebMCP will continue to work with click simulation. But as major players like Expedia and Shopify implement it, AI agents will naturally prefer those structured interfaces — similar to how mobile-first design became essential over time.

Written by Rush

Tracking where business meets AI.

Did you find this reference helpful?

Get curated references delivered to your inbox weekly

Share this reference

Antioch — Meet the Cursor for Robot AI

Physical AI startups no longer need to rent warehouses or build million-dollar test facilities. Antioch brings software-speed development to robotics through cloud simulation — and just raised $8.5M seed to prove it.

Explore more AI workflow guides on similar topics

$20K and 12 AI Tools Built a $1.8B Telehealth Company — And Then the Red Flags Arrived

morningbrew.com

Medvi telehealth, AI startup leverage, GLP-1 startup, one-person unicorn, AI operations

$20K and 12 AI Tools Built a $1.8B Telehealth Company — And Then the Red Flags Arrived

Matthew Gallagher built Medvi, a GLP-1 telehealth startup, in 14 months with $20,000 and AI tools. 2 employees. 16.2% net margin. $401M in year one. Here's how the model works — and where it's breaking.

AI That Works While You Sleep — Automating Recurring Tasks with Claude Code Scheduled Task

substackcdn.com

What if your code review was already done when you woke up, and your newsletter

AI That Works While You Sleep — Automating Recurring Tasks with Claude Code Scheduled Task

What if your code review was already done when you woke up, and your newsletter sources were already organized? Here's how to automate recurring tasks with Claude Code Scheduled Task.

Next →Antioch — Meet the Cursor for Robot AI