AI encoding tools through There is a plan to make the coding agent easier to understand – and it looks terrible, just like paired programming.
Today, the company is releasing Warp Code, a new set of features designed to enable users to monitor more command-based encoding agents, with broader differences tracking, and clearer what the encoding agents do.
“I feel like using these other command line tools, you’re just cross-finger, and hopefully the other end of the agency is something you can actually merge,” founder Zach Lloyd said. With the new features, he wanted to “make a stricter feedback loop for how this agent is coding.”
Actually, this means you can see exactly what the agent is doing and ask questions along the way. “When an agent is writing code, you will be able to see every little difference the agent is making, and you will have an easy way to comment on those differences and adjust as the agent goes.”
Distorted users will be familiar with the general interface: a space at the bottom that provides direct instructions to the agent, and a window to view the agent’s response and side window where you can see the agent’s progressive changes. You can change the code manually, similar to code-based tools like Cursor, but you can also highlight the context of a specific line added as a request or question. Perhaps most impressive is that Warp’s compiler will automatically resolve any errors that occur during compilation.
“It’s about making sure you understand the code the agent is making and make sure it can be edited and reviewed,” said Warp founder Zach Lloyd.
This is a new approach to the increasingly crowded AI-driven programming field. Warp competes with cute completely non-code tools as well as AI-powered code editors like cursors and Windsurf. Foundation Model Company provides its own competition Command Line Tools Just like Anthropic’s Claude Code and Openai’s code, even if Warp uses its model to power its own products.
TechCrunch Events
San Francisco
|
October 27-29, 2025
Warp has 600,000 active users and counts in the AI coding game, but it is still a relatively small player – but it is growing very quickly. Lloyd said the company added $1 million in ARR every 10 days, indicating that there are still many users ready to pay for a better way to code in the atmosphere.
