From Plain Terminal Windows to Intelligent Workspaces
AI-powered terminal tools are terminal emulators and command-line interfaces that embed large language models directly into the command-line workflow, so developers can receive contextual code help, command suggestions, and system insights without leaving the terminal window or switching to a separate AI app. Traditionally, even modern terminals like iTerm2 or Windows Terminal were single-purpose shells: they provided tabs and color schemes, but little awareness of what users were doing beyond displaying text. The rise of AI coding assistant tools began outside the terminal, in browser-based chatbots or IDE plug-ins. Now the trend is reversing. Intelligence is moving into the core developer environment, where command line productivity lives. This shift means the command line is no longer only a place to run commands; it is becoming a live workspace that understands code, files, and processes as users work.
Wave: An AI Terminal Emulator as a Development Desktop
Wave positions itself as more than a traditional AI terminal emulator: it feels like a development desktop that happens to live inside a terminal window. Instead of one big pane and a few tabs, Wave lets users tile multiple views side by side, including a visual file explorer, system resource monitor, web browser, and native GitHub pages. At the center of this layout sits Wave’s AI coding assistant, which can run on local or cloud models, and even an inbuilt model. With the Widget Context option enabled, Wave AI can read what is in all visible panes and answer questions about scripts, system status, or GitHub diffs without extra copy-paste. According to XDA’s coverage of Wave Terminal, this “cyberpunk terminal” setup turns the command line into a place where research, monitoring, and coding coexist in one window.
Command-Line Copilot: How Wave Changes Everyday Tasks
The most striking change with Wave is how it reduces context switching for routine work. Instead of jumping from a browser to a terminal to a documentation tab, users can ask Wave’s AI agent to explain a Python script, outline its limitations, and suggest improvements within the same tiled layout. Features like the wsh ai command make it easy to request one-off explanations or command templates; one XDA writer notes using it repeatedly to remember SSH syntax and have the AI format commands ready to paste into a terminal pane. Wave also adds practical infrastructure features like secrets management, storing SSH passwords, API keys, and tokens through the system’s secure storage, then exposing them via aliases. This shows how AI and core workflow utilities can coexist inside a single terminal tool instead of being split across separate apps.
Crush (formerly OpenCode): AI Coding Sessions in the Shell
While Wave replaces traditional terminal emulators, tools like Crush (the rebranded OpenCode) live as AI coding assistants inside any existing terminal. Crush is a Go-based, open source terminal user interface that focuses on conversational coding sessions with large language models. Users launch it via a CLI command and stay in the shell, where they can plan features, generate code, refactor files, and iterate through prompts. The tool connects to cloud or local LLMs via API keys and allows switching models mid-session, so developers can pair a lightweight model for planning with a stronger one for complex refactors. XDA notes that Crush supports providers such as OpenRouter and Vercel, and can talk to local models through tools like Ollama or LM Studio. The terminal becomes a persistent chat canvas for code rather than a transient prompt history.
Embedded AI as the New Developer Infrastructure Layer
Wave and Crush show a clear pattern: AI help is moving from external websites and IDE plug-ins into the core infrastructure where developers spend most of their time. Wave replaces tools like iTerm2, Windows Terminal, and Konsole with an AI-aware workspace, while Crush brings a full AI coding assistant into any shell session. Users report immediate gains in command line productivity because they no longer need to stop, copy logs, and paste them into a browser chatbot whenever something fails. Instead, the assistant is present at the point of failure, aware of context like active panes or code buffers. This marks a shift from “AI on the side” to “AI in the path,” where intelligent assistance is a default part of terminal workflows rather than an optional add-on, and where the command line starts to feel like a continuous coding copilot.
