A year ago, the AI coding conversation was about getting more AI into your workflow. Now a growing number of developers are asking the opposite: how do I get the autocomplete and nothing else?
The request is specific. Ghost text that appears inline as you type, accepted with Tab, ignored by doing nothing. A chat window where you can ask questions about code. And that is it. No autonomous file editing, no background PR creation, no agent making decisions on your behalf.
Several tools already serve this use case. GitHub Copilot's Vim plugin (copilot.vim) paired with CopilotChat.nvim provides exactly this split: inline completions plus an on-demand chat buffer. Supermaven focuses purely on fast tab completion with a local model option. And minuet-ai.nvim is an open-source alternative that lets you plug in different AI backends while keeping the interaction model simple.
The appeal is control. Agentic coding tools like Cursor's agent mode, Claude Code, or GitHub Copilot's workspace features are powerful, but they change the relationship between developer and editor. Instead of the AI suggesting and the human deciding, the AI starts acting. For developers who have spent years building muscle memory in Vim or Emacs, that shift feels like giving up the thing that makes them fast.
There is also a practical concern. Agentic tools work best with well-structured codebases and clear instructions. In messy legacy code, a vim-with-autocomplete setup might actually be more productive than an agent that confidently refactors the wrong thing.
The tools to watch in this space are Supermaven for pure speed, Copilot for the broadest language support, and Cody for teams that want completion tied to their own codebase context. All three offer non-agentic modes that keep the developer firmly in charge.