I have experimented extensively with Claude Code and Google Antigravity as part of my development workflow. I’ve used them to build an enterprise-grade tool. I wouldn’t count tools alone though. There was a lot of human intervention throughout the process. However, working with them has helped me realize just how capable these tools are.
This has become clearer to me lately Claude Code and Google gave Antigravity a whole project. I wanted to build something useful, not just run another coding tool benchmark. The project was a resume builder microsite for a friend who was actively looking for a job. He already had a solid resume, but he needed a way to edit it properly and create versions tailored to different roles. A single resume rarely works for every application because each job requires a slightly different emphasis.
I wanted the microsite to provide templates, editable sections, and role-specific versions without making the process a manual copy-and-paste. I ran Claude’s Code inside Google Antigravity through the terminal, while also using Antigravity agents to plan and structure the project. This combination finished the build almost twice as fast as my usual VS Code workflow.
Use Antigravity for planning
And Claude’s code for execution
My plan was to use both Google Antigravity and Clau Code to my advantage. I started Antigravity early enough to start planning the project. The plan is to phase the project for Antigravity, create an architecture, and define a database schema. While this particular project doesn’t require an API, if yours does, you can create API properties and ask it to define risks and exceptions. Finally, you can ask it to create executable tasks.
It got things going in a matter of minutes. Then I asked Claude Code to start executing the plan. The best way is to give it executable tasks and stages generated by Antigravity. For my resume builder, this meant tasks like setting up a template system, creating editable sections, and implementing a PDF export. You ask Claude to start working on the first stage and complete the stages created by Antigravity. Claude Code works better instead of a command like “build all” when working on a specific task.
If you want real speed, let Claude Code run on the repository. Claude Code can generate files for you. It can edit existing files, run tests, fix lint errors, and search the entire repository. VS Code, on the other hand, asks you to manually switch between tabs. For example, when working on a resume builder, I could ask Claude Code to update each template to support a new field or change the PDF export flow. It can work and complete a task without opening dozens of files.
Both tools have their strengths
Once you get a reasonable result from Claude Code, you can use Antigravity as a second opinion. You can ask Claude Code to do something, then ask Antigravity to review it, and then feed the feedback back to Claude Code. Antigravity can identify edge cases, usability issues, or architectural issues. Once these issues are highlighted, you can propose the Claude Code to implement the changes.
Although my resume builder project didn’t require both tools to run over time, if you’re working on something bigger, you can also split the work into parallel streams. Instead of instructing Claude Code to build the backend, waiting, then asking it to build the frontend, waiting again, and asking it to write documentation, you can run multiple sessions. In my case, Session A can focus on setting up authentication and Session B can focus on the template editor. At the same time, the C session can handle the creation or documentation of the PDF. The idea is to use multiple agents so that different parts of the project can progress in parallel.
I also found it useful to split the implementation and reasoning between the two tools. I found Antigravity to be better with architectural planning, design decisions, and overall case review. Claude Code excels at writing code, refactoring code, executing commands, and creating tests, so you can split these functions between tools for better results.
VS Code feels behind in agent development
It lags far behind at this point
VS Code still has a huge ecosystemand I understand why developers use it. It gives you extensions, debugging, Git support, and pretty much every workflow a developer would expect. Microsoft also recently added AI features and Copilot-based agent workflows to VS Code.
But my problem with VS Code comes from how the workflow feels during a full project. You can add extensions, open terminals, install Copilot, configure tools and build a decent installation. This setup still depends on you linking the workflow together.
Since the official product Code is different from the OSS repository, VS Code also carries baggage. Microsoft develops Code OSS under the MIT license, but Visual Studio Code ships as a Microsoft product with Microsoft-specific customizations under the traditional Microsoft product license. It also explains why tools like VSCodium exist for people who want cleaner open source binaries. Telemetry also adds to my frustration with VS Code.
Other tools now feel more aggressive about the future of coding. Cursor, Gravity, Zed, Warp and similar tools focus more on agent workflow. VS Code still gives you the most features through extensions and integrations, but that doesn’t automatically make it the fastest workflow.
Using multiple agents is the way to go
We have come to such a point using one agent for everything probably not the best approach anymore. You can use multiple agents in the same project and get more value from them. I also found it to be cheaper than running everything through an agent. This allows cheaper models to be used for simpler tasks, while keeping the more capable models for the work that needs them.
