IrisGo: Andrew Ng-Backed Startup Builds an AI Desktop Buddy That Learns Your Workflow

Industry insiders have long predicted that the next frontier in artificial intelligence is the rise of proactive systems: AI agents that don’t just wait for commands but anticipate a user’s needs and act on them autonomously. A new startup, IrisGo, is positioning itself at the forefront of this shift with a desktop companion designed to learn and automate daily workflows with minimal human prompting.

Founded by Jeffrey Lai, a former Apple engineer who helped build the Chinese language version of Siri, IrisGo recently closed a $2.8 million seed round led by Andrew Ng’s AI Fund. The company’s name is a deliberate inversion of Apple’s assistant, hinting at its ambition to reimagine the concept. The core idea is straightforward: a user demonstrates a task once, and IrisGo records the process, remembers it, and can execute it autonomously in the future.

From Coffee Orders to Complex Workflows

During a recent demonstration with Bitcoin World, Lai showed how IrisGo could learn to place a coffee order online. The agent recorded the steps—selecting a latte from Philz Coffee, entering credit card details, and hitting purchase—and then repeated the entire process without further instruction. While the example is simple, the underlying technology is designed for far more complex applications.

IrisGo comes with a built-in library of pre-built skills, including email drafting, invoice processing, report generation, and document summarization. Beyond these pre-configured tasks, the system continuously learns from the user’s desktop behavior, adding new actions to its repertoire. The application also includes a coding assistant similar in concept to OpenAI’s Codex or Anthropic’s Claude Code, aimed at helping developers automate routine programming tasks.

Targeting the White-Collar Productivity Gap

Lai describes the target audience as knowledge workers in white-collar industries. Despite the rapid advancement of large language models, many office tasks remain surprisingly manual and repetitive. The goal is to free workers from clerical drudgery, allowing them to focus on higher-level conceptual work while the AI handles background operations.

This focus on practical, everyday productivity distinguishes IrisGo from many AI tools that emphasize raw capability over seamless integration into existing workflows. The company’s strategy is to embed itself into the daily routines of office workers, becoming an invisible but indispensable layer of automation.

Privacy as a Differentiator

A key selling point for IrisGo is its emphasis on on-device processing. The system is designed to handle a significant amount of data locally, offering stronger privacy protections than cloud-dependent alternatives. Lai acknowledges that the architecture is hybrid; more complex tasks may still require cloud processing, but the company states that this only occurs with explicit user authorization and uses end-to-end encryption. For businesses concerned about data security, this local-first approach could be a decisive advantage.

Backing from Tech Heavyweights

IrisGo has secured credibility through association with prominent figures. Andrew Ng, a co-founder of Google Brain, led the seed round through his AI Fund after Lai and his co-founder, both Carnegie Mellon alumni, secured a meeting through a shared connection. Nvidia and Google have also invested in the company. This level of backing not only provides capital but also signals confidence in the technology’s potential.

The startup recently launched beta versions of its macOS and Windows apps and is actively pursuing deals with laptop manufacturers to preinstall the software on new devices. A deal with Acer has already been struck, and Lai indicated that similar agreements with other OEMs are in the pipeline.

Why This Matters

The rise of proactive AI agents represents a fundamental shift in how we interact with computers. Instead of being a passive tool that responds to commands, the computer becomes an active participant in the user’s workflow. If IrisGo succeeds, it could redefine productivity software, making automation accessible to millions of office workers without requiring technical expertise. However, the company faces significant challenges, including user trust, privacy concerns, and the technical difficulty of reliably automating diverse and unpredictable human workflows.

Conclusion

IrisGo is an ambitious attempt to bring proactive AI to the desktop, backed by some of the most respected names in the industry. Its focus on learning by demonstration, on-device privacy, and practical business tasks positions it as a serious contender in the emerging market for autonomous AI agents. Whether it can live up to its promise will depend on execution, user adoption, and the ability to navigate the complex landscape of enterprise software.

FAQs

Q1: What exactly does IrisGo do?
IrisGo is a desktop application that learns how a user performs repetitive tasks on their computer and then automates those tasks. It can be shown a process once and will remember it for future use.

Q2: How is IrisGo different from other AI assistants like Siri or ChatGPT?
Unlike traditional assistants that require explicit commands, IrisGo is designed to be proactive. It learns from user behavior and can suggest or execute automations without being asked, aiming to anticipate needs rather than just respond to them.

Q3: Is my data safe with IrisGo?
The company emphasizes a privacy-first approach, processing much of the data on the user’s device. Cloud processing is used only for complex tasks and requires explicit user permission, with end-to-end encryption in place.