I'm Aloke, first engineer and head of the coding team at Warp. AMA! 🔥

Really glad to see how Warp keeps evolving, especially the new feature for opening files and adding context! I had a few questions:

• In one of the blog posts, it says that requests are basically counted by tokens rather than the number of user prompts. Is there any plan to make the pricing model more transparent or simplify it, similar to Windsurf’s approach where usage is just based on the user’s prompt instead of each tool call/operation? In a 2–3 hour session I’ve noticed I can easily use up 200–300 requests on a handful of small to medium-sized tasks, which adds up pretty quickly.

• Are there any plans to support native web search without having to rely on an MCP server?

• Right now I’m still using Cursor for things like autocomplete and seeing git diffs. Do you see Warp eventually including some of those IDE-style features? I know that might not align with the product philosophy of keeping Warp more terminal-focused, but I’m curious if certain core features might make it in so I can lean less on other tools over time.

Keep up the great work!

For folks not used to working in a terminal what does the common terminal setup look like for you and your team now that you are using/making warp, or has your usage pretty much been consumed by warp?

Hi Aloke,

Thanks for doing this AMA. I'm a huge fan of Warp's core terminal experience, but I'm struggling to understand the strategic positioning of the new coding features.

For me, modern AI agents are pushing my workflow more into my IDE (JetBrains/Vim), not less. The terminal is for execution, and the IDE is for complex editing and navigation. Warp is trying to build editing features, but it's a long way from competing with the mature ecosystems of JetBrains or Vim.

It feels like you're building features that already exist in the tools where developers spend most of their time. I have a hard time fitting Warp into my serious dev work because I live in my IDE.

As one of your biggest fans who currently finds the product hard to use for this reason, my question is: have you considered focusing on deeper integration with IDEs instead of trying to replicate them? Making Warp the best possible terminal inside of VS Code or a JetBrains IDE seems like a much more powerful value proposition.

I've love to see soon a way to schedule agents and also orchestrate complex agent workflows(node based preferably), a easy and customizable way of calling agents and schedule them at specified intervals, or for a small agent to more frequently read data and call the main actor.

Love the direction Warp is taking with agentic coding. The balance between giving AI autonomy and keeping developers in the loop feels like the sweet spot — agents that augment, not replace. Curious how you see this evolving for larger teams where collaboration and consistency become harder to manage. We’ve been exploring similar ideas with AI agents inside Growstack, and team workflows are where things get really interesting.

Congrats on Warp Code. The marketing team is killing it!

I know Warp is an ADE and this might be redundant, but what if I wanted to use Warp inside of say Rider when I'm working on my game? Right now, you can do that with Claude Code. Maybe this defeats the purpose of Warp being an ADE, but is there any plan for this use case in the future?

Congrats, Aloke. Really cool to see tools that help developers work with agents instead of just replacing them. I’m curious , which of these new features do you think will make the biggest difference day-to-day for devs?

Love seeing Warp double down on agentic coding while keeping humans in the loop. The new code review panel, file editor, and Warp md setup make it feel like pair programming with superpowers. Excited to see how this shapes the future of dev workflows.

HI Aloke! Good luck with WARP.md , btw is there english version for that? people can translate it using google but.... just to make people easier when clicking your product to sight seeing