This Desktop AI Agent Watches You Work Once — Then Does It for You Automatically

IrisGo has raised $2.8 million in seed funding led by Andrew Ng's AI Fund, with participation from Nvidia and Google, to build a desktop AI agent that learns and automates user workflows without requiring repeated instructions. The company was co-founded by Jeffrey Lai, a former Apple engineer who worked on the Chinese-language version of Siri.

The core premise of IrisGo is that most knowledge workers repeat the same sequences of steps daily — placing orders, processing invoices, drafting and routing emails — and that an agent capable of observing those workflows once should be able to handle them autonomously from that point forward. The platform watches a user complete a task, builds an internal model of the workflow, and then executes it independently in the background going forward. A built-in skills library covers common tasks out of the box, while the agent continuously expands its repertoire from observed desktop behavior.

Lai described the goal as moving users away from the current model of AI interaction, where humans repeatedly prompt a model to complete individual steps, toward a mode where autonomous agents handle routine work entirely, freeing people for higher-level decisions. A coding assistant described as comparable to Claude Code is also included in the platform.

Privacy is a central part of IrisGo's design pitch. The platform uses a hybrid architecture that processes as much as possible on-device, with cloud-based processing available only when the user explicitly authorizes it. This positions IrisGo as a more privacy-conscious alternative to purely cloud-based desktop agents, a consideration likely to matter for enterprise buyers handling sensitive workflows.

Beta versions are live for both macOS and Windows. IrisGo has already secured a preinstallation deal with Acer and is in discussions with other device manufacturers about similar arrangements — a distribution strategy that, if successful, would put the agent in front of users before they ever visit an app store.

The round adds IrisGo to a crowded but rapidly expanding field of agentic desktop tools. Competitors range from large incumbents building agent layers into existing platforms to a wave of well-funded startups targeting the same knowledge-worker automation market. The combination of Andrew Ng's backing, hardware preinstall deals, and a privacy-forward architecture gives IrisGo a distinct positioning as the category grows.

This analysis is based on reporting from The AI Insider.

This article was generated with AI assistance and reviewed for accuracy and quality.