Claude and Gemini Can Now Work Together to Build Your Business Apps

Google Cloud said on 20 October 2026 that its AI coding platform, Antigravity, now lets developers choose Anthropic's Claude Opus 5.5 or Claude Sonnet 5.5 alongside Google's own Gemini models in the same project. The update means one AI model can review and fix another model's code before it ships.

What did Google announce?

Antigravity is Google's AI-powered development environment, built around Antigravity 2.0, the Antigravity CLI, and IDE extensions for VS Code, JetBrains and Visual Studio. According to Google Cloud's 20 October 2026 announcement, Claude Opus 5.5 handles complex design decisions while the lighter Claude Sonnet 5.5 runs routine edits faster and cheaper, and both now sit next to Gemini 3.8 Flash and Gemini 4 Argon in the same model picker. An administrator switches the models on once, then every developer on the team can pick whichever model fits the task, under one bill and one security policy.

At its core, Gemini is all about giving your organization an open, reliable platform where builders have choice and teams work with complete confidence.from Jamie de Guerre, Senior Director, Outbound Product Management, Google Cloud (Google Cloud Blog, 19 October 2026)

What can Claude and Gemini do together now?

• Cross-model verification: one model writes or edits the code, a different model family checks it before the change goes live, catching mistakes a single model might miss on its own.
• Reasoning plus fast subagents: a slower, more careful model plans the work, then quick helper models handle repetitive jobs like running tests or sorting through error messages.
• Mid-project switching: a developer can swap models between messages in the same conversation without losing the files, notes or terminal history already built up.
• One bill, one admin panel: whoever manages the account turns on Claude models once, and the cost shows up in the same Google Cloud invoice as everything else.

How could your business use this?

You do not need a developer on staff to benefit from this shift. A clinic's front-desk team could have someone vibe code a simple booking form, with one AI model drafting the page and a second model proofreading it for bugs before patients ever see it. A retailer tracking stock across two shops could build a small inventory tool the same way, catching a costly formula error because a second model flagged it first. The point is not which AI brand wins. It is that checking AI-written work with a second, independent AI model is now a built-in option, not a manual step you have to remember.

Also today

• Anthropic said its newer safeguards have cut the data copied out of Claude through distillation attacks by 90 to 99 percent, and the company has banned tens of thousands of accounts caught trying to clone the model's responses (Anthropic).
• Google Cloud was named a Leader in the first IDC MarketScape ranking of worldwide agentic AI platforms, with Gemini's reach across builder and business tools cited as a deciding factor (Google Cloud Blog).
• Malaysia's Budget 2027 sets aside RM30 million through MDEC to help 4,000 micro, small and medium businesses adopt AI and automate their operations, part of a push to train 200,000 AI-skilled workers by 2030 (Malay Mail).

What to do this week

If your business already has someone experimenting with AI-built tools, point them at a platform that lets two models check each other's work, the way Antigravity now does. It costs you nothing to ask a model to review its own code with a second model before anyone else touches it. In our workshops, we tell business owners the same thing about any AI output: never ship the first draft unchecked, whether that is a customer email, an invoice summary or a line of code. If you want a guided way to turn your own process into a working app without hiring a developer, NextGen Training Academy's AI Full-Stack Building for Business programme in Kota Kinabalu walks you through exactly that, from a blank page to a tool your team actually uses.