"We chose to work with Claude on Vertex AI thanks to Claude's instruction-following, tool-calling execution, and reliable code generation. It offered incredible performance," says Gilboa.

As the AI works, it "explains" to the user what it's doing, including how it's creating the tool architecture. This transparent process educates users and fosters the trust they need to play with ideas.

Users can also change or refine the AI's output by using prompts and selecting integrations to add features. For example, if they want to add a chatbot to a dashboard, they might prompt spring.new to "add a chatbot that can answer with real data from my dashboard using …" and then click to integrate a foundation model available on spring.new like Claude or Gemini.

"Typically, people aren't going to know exactly what they want on the first prompt," says Gilboa. "The fact that they can iterate with spring.new is really special. Vertex AI and Claude make this possible."

Agentic capabilities eliminate the need to write a completely new app before testing the logic. spring.new's agent can use different integrations, understand if something is wrong with the way it thought about the problem, and then update the app when it reasons it's ready.

"With Google Cloud and Anthropic, spring.new is redefining how we work," says Hizmi.

Just a couple of years ago, development of business applications could take weeks to months and required technical support. Now, anyone can start with a simple one-sentence prompt to get a result in minutes, and then iterate, building layer upon layer of functionality until they've successfully created what they need to do their best work.

"It's very exciting," adds Gilboa. "With the compounding benefits of using Google Cloud's infrastructure and Anthropic's ever-evolving models, combined with ever-more capabilities added to the spring.new platform, we will see many more teams tapping into their creative potentials — to work in ways that were never possible before."