Built for operations staff and non-technical professionals doing repetitive desktop work, Lapu AI replaces manual file sorting, PDF data pulls, and multi-app busywork with an agent that acts instead of only advising. It can reorganize hundreds of files in one pass and logs every step so users can review exactly what ran before trusting it with more.
Founded in 2025, Lapu AI is a native desktop agent for macOS and Windows that executes file, shell, and app tasks directly on a user's machine rather than only suggesting them. Its defining trait is mandatory per-action approval before any write, shell command, or state-changing click runs, with low-risk reads handled automatically.
Maker: Lapu AI · Autonomy: semi autonomous · Maturity: GA
About Lapu AI
Lapu AI is a desktop AI agent that runs natively on macOS and Windows, giving it direct access to the file system, terminal, and desktop applications instead of only suggesting actions like a browser-based assistant. Founded in 2025, it entered a market split between passive screen readers such as Microsoft Recall and active execution agents, and positions itself squarely in the latter camp.
The agent pairs a built-in frontier AI model with local tools: file read, write, edit, and grep; shell command execution; and OS-level accessibility APIs that let it click, type, and navigate inside desktop apps it was never designed to control. Every sensitive action, such as a file write, a shell command, or a state-changing click, pauses for an explicit per-action approval prompt, while low-risk reads run automatically.
Typical workflows include document processing, data extraction from PDFs and spreadsheets, file organization across hundreds of items, research synthesis, inbox triage, and recurring multi-step reports for operations teams that would otherwise do this by hand.
Pricing
Free plan ($0/mo, no credit card): full desktop app, frontier AI within monthly allowance, all core tools. Premium $20/mo, Pro $60/mo, Max $100/mo raise the AI reasoning allowance. Team and enterprise tiers add SSO/SAML and shared usage pools.
Key Features
- Permission-Gated Execution: Every file write, shell command, and desktop input pauses for explicit user approval before running, with low-risk reads auto-approved and a 90-day audit log of every action taken.
- File and Shell Automation: Organizes, renames, and searches thousands of files by type, date, or content, and runs terminal commands with chained operations without the user needing to remember syntax.
- Desktop Automation via Accessibility APIs: Controls any macOS or Windows app using OS-level accessibility APIs to list windows, take screenshots, click, type, and navigate inside apps the agent was not originally designed for.
- Reusable Skills and Scheduling: Package any multi-step workflow into a named skill that can be rerun or scheduled to trigger automatically, turning one-off tasks into recurring operations without re-prompting.
- Local-First Privacy: All files and data stay on the user's machine with no Lapu AI cloud storage or remote workspace; only the context needed for AI reasoning is sent to the model endpoint during a task.
- Mini Chat Overlay: A persistent overlay lets you describe a task from any app without switching windows, so the agent starts working without interrupting what you already have open on screen.
Strengths
- Includes the full desktop application and every core tool at no cost, unlike most desktop agents that require a subscription from day one.
- Gives more visibility into what the agent actually did than cloud-based agents, where actions run in a remote sandbox you cannot inspect directly.
- Runs on both macOS (Apple Silicon) and Windows (64-bit), covering the two dominant desktop platforms without feature parity gaps between them.
Weaknesses
- The underlying frontier model powering the agent is not publicly named, making it impossible to compare reasoning quality against agents with published benchmark scores.
- No published benchmark scores (SWE-bench, WebArena, GAIA) as of June 2026, so objective performance comparisons with agents that do publish results are not possible.
- macOS support requires version 12 Monterey or later on Apple Silicon, which excludes older Intel Macs and any machine below that OS version.
Frequently Asked Questions
What are Lapu AI's pricing plans in 2026?
Lapu AI offers four tiers: a free plan with no credit card required, Premium at $20/mo, Pro at $60/mo, and Max at $100/mo. The free tier includes the full desktop app and all core tools within a monthly AI usage allowance, and paid tiers raise that allowance. Team and enterprise plans add SSO/SAML authentication and shared usage pools, with pricing available on request.
What do you get on Lapu AI's free tier?
The free plan requires no credit card and includes the full desktop application, all core tools, and a monthly AI reasoning allowance sufficient for dozens of typical workflows. When that allowance runs out, the agent stops with a clear message rather than degrading or charging automatically. Upgrading to a paid tier only changes the size of the allowance, not which features are available.
What should you use instead of Lapu AI?
Claude Code is the closest comparison for developers, since it is a terminal-based coding agent rather than a GUI desktop automator. Eigent is a free, open-source multi-agent desktop that runs several parallel local workers instead of Lapu AI's single gated agent. Microsoft Copilot on Copilot+ PCs offers passive, suggestion-only assistance built into Windows rather than an agent that executes tasks end to end.
Is Lapu AI better than Claude Code?
Claude Code is a command-line coding agent scored at 80.8% on SWE-bench Verified, aimed at developers who want a verified coding benchmark result. Lapu AI has no published benchmark scores, but its graphical desktop interface automates any app via accessibility APIs rather than only a terminal. Developers doing heavy coding work are better served by Claude Code, while anyone whose work spans file management, PDFs, and everyday desktop apps gets more from Lapu AI.
How do you set up Lapu AI?
Download the desktop app from lapu.ai for macOS 12 Monterey or later on Apple Silicon, or Windows 10/11 64-bit, and create a free account with no credit card. A setup walkthrough requests the accessibility permissions the agent needs to control your desktop, after which you can type your first task into the mini chat overlay, such as organizing PDFs in a folder by date. The agent proposes its steps and waits for approval on anything that writes, deletes, or runs a command.
Top Alternatives
- Claude Code: Claude Code fits developers chasing a verified coding benchmark from a terminal; Lapu AI fits better when the work is desktop GUI automation across everyday apps rather than a codebase.
- Eigent: Eigent suits users who want a free, open-source multi-agent desktop running several parallel local workers; Lapu AI suits those who prefer one agent gated by mandatory per-action approval.
- Microsoft Copilot: Microsoft Copilot works as an AI companion built into Windows and Office for everyday productivity; Lapu AI goes further, executing multi-step file and shell tasks with explicit approval instead of only assisting inline.