AI has made it easy to generate software code, but some open source projects have stopped taking code submissions from the public, citing a deluge of low quality code or code that doesn’t match project needs.
Warp, maker of tools for AI coding, is moving in the opposite direction. It’s making its desktop agentic development environment (ADE) software open source and even encouraging users to contribute new features with the help of AI. The ADE lets humans and AI agents work together to write code.
Founder and CEO Zach Lloyd says software developers typically have their own preferences on tools and working styles, and he anticipates the program will let some of its nearly 1 million users add new features that best suit their needs faster than the Warp team could do on its own.
“One thing that I’ve realized the longer I’ve been building for developers is that they have very diverse workflows,” Lloyd says. “So the way that we can get Warp to be the most powerful product for the greatest number of developers is to let those developers build the pieces that our internal team probably doesn’t have the bandwidth to do.”
Still, Lloyd doesn’t anticipate simply accepting bundles of code from strangers on the internet. Instead, developers interested in adding a new feature or addressing an issue in the code can propose a plan on Warp’s GitHub issues page. Warp’s AI agents will then take a first stab at responding, including asking for more details on what a developer is looking to do and potentially generating a fleshed-out specification. A human at Warp will ultimately make the call on whether it’s a change the company would want to see in the app.
“The idea is that agents do a bunch of the grunt work around triage and spec’cing out initial ideas, but humans are kind of still in the loop deciding what to build, and providing instruction on how to do it,” says Safia Abdalla, a software engineer at Warp.
Once an idea is approved, developers will be able to get to work and, if they wish, they can use Warp’s own Oz orchestration software to manage agents building the code in the cloud. At least at the start, Warp will pay for Oz usage and AI credits to do so, Lloyd says. If developers prefer, they can also work on their own computers with tools of their choice and submit a pull request through GitHub when the work is done and ready for review by AI agents and humans at Warp.
“We will ultimately code-review everything and make sure that the stuff that gets merged is high quality,” he says.
Not all of Warp’s software will be open source: The cloud-based elements that make up much of Warp’s enterprise business will still be proprietary, at least for the time being, and there may be elements of the desktop app designed for specific clients or used to test not-yet-available AI models that won’t be released to the public, Lloyd says.
And with the source available, users will be able to make their own changes to the software for their own use without submitting it back to Warp. “I think we’re in an age where so many people can build software, they can take forks, they can make it work for them how they want it to work,” Lloyd says. “If they want to do that, that’s totally fine.”
Warp has already evolved rapidly, starting from an AI-empowered version of the command line terminals programmers use for precision control of their computers and evolving to add a coding environment designed for collaboration with AI agents. That, in turn, was followed by Oz, its cloud-based AI orchestration software.
In theory, with the desktop software open source, someone could create their own version of the development environment and attempt to compete with Warp. But Lloyd says he’s not too concerned about the risk, especially since the software license will prevent anyone from distributing a closed-source version.
And in addition to helping the product better match the needs of its users, going open source will also let Warp offer a public demonstration of what’s possible with its AI tools.
“It’ll help our enterprise business by being a sort of showcase app for what an agent-powered software project can be like,” Lloyd says.