The next frontier in artificial intelligence is moving beyond reactive chatbots and into the realm of proactive systems — agents that can anticipate a user’s needs and fulfill them before the user even realizes what those needs are. Industry insiders have been buzzing about this shift for months, and one startup is positioning itself at the forefront: IrisGo. The company, which secured a $2.8 million seed round earlier this year led by Andrew Ng’s AI Fund, is building a desktop companion for PCs that learns from daily workflows and automates them with minimal or no human prompting.

The Vision Behind IrisGo

IrisGo was founded by Jeffrey Lai, a former Apple engineer who contributed to the Chinese language version of Siri. The name Iris is a sly reversal of Siri, hinting at both its lineage and its aspirational role. The core idea is deceptively simple: show the program how to perform a task once, and it remembers the entire sequence of steps for future automated execution. No repeated instructions, no complex configuration — just a demonstration and then autonomous repetition.

During a demo with Lai, the platform recorded the process of placing a coffee order online: selecting a latte from Philz Coffee, filling in credit card details, and hitting the purchase button. He then instructed IrisGo to repeat the order on its own, and the agent dutifully complied. While buying coffee might seem trivial, the underlying capability has profound implications for business productivity. IrisGo is designed to handle a wide variety of repetitive clerical tasks, freeing up knowledge workers to focus on higher-level conceptual work.

Core Features and Capabilities

The application comes with a built-in skills library that includes prebuilt automations for email drafting, invoice processing, report generation, document summarization, and many other common office workflows. Beyond this library, IrisGo continuously learns from a user’s desktop behavior, automatically identifying patterns and suggesting new task sequences that could be automated. This creates an ever-expanding set of action items tailored to each individual’s work habits.

Additionally, IrisGo includes a coding assistant. This feature operates similarly to tools like OpenAI’s Codex or Anthropic’s Claude Code, providing real-time support for developers as they write, debug, and refactor code. The coding assistant is context-aware, understanding the broader project environment to offer relevant suggestions.

Target Audience and Use Cases

According to Lai, the primary audience for IrisGo is knowledge workers — white-collar professionals who spend a significant portion of their day on repetitive tasks. Despite the advances in frontier AI models, many office workers still engage in manual, time-consuming processes. IrisGo aims to bridge this gap by enabling more fully autonomous workflows. The human handles strategic decision-making and creative problem-solving, while the AI takes care of the clerical drudgery in the background.

The platform has already entered beta on both macOS and Windows. The company is also actively pursuing preinstallation deals with laptop manufacturers. A deal with Acer has already been struck, and Lai expressed hope that similar agreements with other device makers would follow. This strategy mirrors the approach used by early virtual assistants like Cortana and Siri, which were bundled with hardware to gain widespread adoption.

Privacy and Architecture

One of the standout features of IrisGo is its emphasis on privacy. A significant portion of data processing occurs on the user’s device, reducing the need to send sensitive information to the cloud. However, the system employs a hybrid architecture: larger, more complex tasks are processed through cloud servers, but the company promises that cloud processing only occurs when explicitly authorized by the user and uses end-to-end encryption. This approach balances the computational demands of advanced AI with the privacy expectations of users, particularly in enterprise environments.

Lai emphasized that the hybrid model allows IrisGo to leverage the power of cloud-based models when needed while maintaining user control over sensitive data. This is increasingly important as regulations like GDPR and CCPA place stricter requirements on data handling.

Backing from Industry Titans

The startup’s credibility has been significantly boosted by its association with prominent figures in the AI world. Andrew Ng, co-founder of the Google Brain deep learning research team and a renowned educator, led the seed round through his AI Fund. Ng’s involvement not only provided capital but also validated the technological direction of the company. Lai connected with Ng through a shared alma mater, Carnegie Mellon University. During an initial meeting, Lai and his co-founder demoed IrisGo, and Ng’s fund decided to invest.

Beyond Ng, IrisGo has also received backing from Nvidia and Google. These investments signal confidence from some of the biggest players in the AI ecosystem. Nvidia, in particular, has a vested interest in platforms that can drive demand for its hardware, while Google’s involvement suggests a strategic alignment with its AI development efforts.

Expanding the Landscape of Proactive AI

The concept of proactive AI is not entirely new. Virtual assistants like Siri, Alexa, and Google Assistant have long attempted to anticipate user needs, but their capabilities have been limited. IrisGo differentiates itself by focusing on complex, multi-step workflows that require memory of specific sequences. It is less about answering queries and more about executing tasks autonomously.

This approach aligns with the growing trend of agentic AI, where systems are designed to take actions in the world rather than merely provide information. Competitors in this space include startups building AI agents for customer service, data entry, and software development, but IrisGo’s focus on the desktop environment gives it a unique position. By embedding itself into the daily workflow of knowledge workers, it has the potential to become an indispensable tool.

Historical Context and Future Outlook

The development of IrisGo can be traced back to earlier efforts to automate office tasks, from macros in spreadsheet software to robotic process automation (RPA). However, those tools required explicit rules and programming knowledge. IrisGo leverages large language models and reinforcement learning to adapt to new tasks with minimal supervision. This represents a significant leap forward in usability.

The company’s ambition does not stop at individual productivity. By preloading IrisGo on new laptops, it could become a default component of the computing experience. As more users rely on it for everyday tasks, the system will accumulate a vast repository of common workflows, further improving its ability to anticipate needs. Over time, the line between human intention and machine execution may blur, reshaping how we interact with our devices.

Lai noted that despite the power of today’s frontier models, AI-assisted office work can still feel incredibly manual and repetitive. IrisGo aims to change that by making the AI invisible and effortless. The system’s ability to learn by demonstration is reminiscent of programming by example, a long-standing goal in human-computer interaction. If successful, IrisGo could transform the desktop from a passive tool into an active partner.

The beta versions of the macOS and Windows apps are now available, and the company is actively onboarding users. With strong financial backing and partnerships with major hardware and AI companies, IrisGo is well-positioned to become a key player in the next wave of AI-driven productivity tools.

Source: TechCrunch News