Lumichats

About Lumichats

Lumichats is a desktop application that lets you run an AI agent on your machine using plain English, without touching a terminal. It reads your local files, executes commands, and produces finished documents like .docx or .pptx, while charging only for the work you run, with no subscription. The tool runs its model on remote servers, and the local app enforces a permission gate for every file write, edit, and command.

Review

Lumichats targets people who want AI agents to do real file work on their computer but find terminal-based tools inaccessible. The app turns natural language requests into local actions-the model proposes a command, the app asks for approval, and then executes it. This review looks at what the current Windows release actually does, its pricing model, and where it falls short.

Key Features

• Sends model responses as tool calls to your local machine, where the app executes them after permission. The model itself never accesses your filesystem directly.
• Asks for consent before changes with three modes: ask before each action, auto-apply (with scoped folder grants), or read-only. Destructive commands can never be pre-approved.
• Snapshots every file it writes or patches right before the change, so you can rewind to the exact previous bytes. Script-generated files aren't covered by this rollback.
• Logs every source the model reads, including the query used and whether the page was actually opened or only appeared in search results-so you can see what was actually retrieved.
• Connects to any MCP server, letting you point it at databases, issue trackers, or internal search tools.

Pricing and Value

Lumichats uses a usage-based model with no subscription. The developer describes a day pass that gives unlimited use for that day, costing under a dollar for an afternoon of work. You pay only when you use it; nothing runs in the background, so there are no ongoing charges. Exact pricing tiers beyond the day pass are not yet detailed in the available information.

Pros

• Translates each command into a plain-language sentence derived from the actual tool arguments, so you see "Create portfolio.html" with the exact path beneath it, not a raw command.
• Permission scoping lets you say "always allow writes in this folder" for a session, reducing approval fatigue without bypassing destructive-action guards.
• Built-in rollback restores files to their pre-write state, covering the common case where the agent edits multiple files and you don't like the result.
• Runs on Windows now, with macOS and Linux builds planned but not yet released according to the roadmap.

Cons

• The Windows installer is not code-signed yet, so SmartScreen shows a warning. A SHA-256 hash is published for verification, and a certificate is planned.
• Conversation data and the content of files you point the agent at are sent to the remote model server, which changes the privacy posture compared to a fully offline tool.
• Not well suited for users who require that all processing stay on their own machine-the separate LumiChats Offline product is built for that use case, but this desktop app relies on cloud-based models.

Lumichats fits professionals who need an agent to operate on local files, generate documents, and run commands without learning a CLI, and who are comfortable with the remote model's data flow. The day pass pricing removes the pressure of a per-action meter, though the lack of code signing and the Windows-only launch limit its immediate reach. If you're handling sensitive source material that can't leave your device, the offline sibling product remains the more appropriate choice.

Open 'Lumichats' Website