Warp

Warp is an agentic development environment built around a modern, GPU-accelerated terminal and a multi-agent orchestration layer for software engineering tasks. It started life as a rewritten terminal emulator with block-based command output, autocomplete, shared workflows and team-oriented features, and has since evolved into a full agentic coding platform where developers delegate work to AI agents that plan, run commands, edit files, review diffs and iterate on their own. The Warp Agent runs locally against your shell and repository, while the Oz cloud agent platform lets you launch and manage long-running agents from a browser, CI or the CLI. Under the hood Warp is model-agnostic: it routes tasks across frontier models from OpenAI, Anthropic (Claude), Amazon Bedrock, OpenRouter and LiteLLM, with codebase indexing, permissioned tool use and per-command approval controls. It suits individual developers who want a smarter terminal with an inline copilot, as well as platform and infrastructure teams that need governed, auditable agent workflows across many repos. Common uses include debugging failing builds, writing and running shell one-liners, scaffolding features from a natural-language brief, doing agentic code review on a PR, running incident response playbooks, migrating legacy code, refactoring across a monorepo, and orchestrating multiple parallel cloud agents on backlog tickets. Business and Enterprise tiers add SAML SSO, admin data-control policies, team usage metrics, custom credit pools, bring-your-own-LLM and self-hosted cloud agents for regulated environments. The core Warp terminal client is open source, which is unusual for a commercial AI coding product and makes it easier to adopt inside security-conscious organisations.