Andrew Ng, the co-founder of Coursera and former head of AI at Baidu, is backing IrisGo, a desktop AI assistant startup that just raised $2.8 million in seed funding. The company's core innovation is simple but significant: it processes user data on your computer rather than sending it to distant cloud servers, addressing privacy concerns that have plagued mainstream AI assistants like Microsoft Copilot and Google's AI Overviews.

IrisGo positions itself as an "AI butler" that sits quietly in your system tray, intercepts keystrokes, and automates repetitive desktop workflows without exposing sensitive information to external servers. Founder Jeffrey Lai's vision reflects a growing consumer preference for what technologists call "offline-first" tools, a trend reinforced by Apple's emphasis on on-device AI in iOS 18.

Why Does Privacy Matter More Now Than Ever?

The privacy angle is not merely marketing rhetoric. According to a 2025 survey by Deloitte, 61% of American consumers now prefer AI tools that process data on-device, up from 42% in 2023. This 19-percentage-point shift in just two years signals a fundamental change in how people think about AI assistants. The timing is crucial: the European Union's Artificial Intelligence Act (AI Act) entered into force just weeks before IrisGo's seed round was disclosed on May 20, 2026, imposing stricter rules on cloud-based AI services.

By positioning itself as a privacy-compliant alternative, IrisGo may tap into what investors call regulatory arbitrage, potentially accelerating adoption across Europe and other privacy-sensitive markets. However, there is a trade-off: local models are often less capable than their cloud counterparts, which could limit IrisGo's appeal to power users who need maximum performance.

What Does Andrew Ng's Involvement Signal About AI Investment Right Now?

Ng's track record speaks for itself. His early bets on deep learning have repeatedly translated into market-leading platforms, from Google Brain to Landing AI. His decision to back IrisGo suggests he sees a structural gap in the market: existing AI assistants are either too generic or too enterprise-focused, leaving a void for a desktop-native tool that feels both personal and professional.

Yet Ng's involvement carries risk as well. His reputation is tied to the deep learning paradigm that has lately faced scrutiny over cost and energy consumption. If the technology stumbles, investor sentiment could shift rapidly.

The composition of IrisGo's seed syndicate reveals broader trends in AI funding. While Ng's presence provides credibility, the round also includes angels and micro-venture capital firms rather than giant growth funds, signaling a preference for capital efficiency over splashy valuations. This cautious stance mirrors the broader tech sector. Crunchbase data show that AI seed rounds in the United States fell 40% in the first quarter of 2026 compared with the same period the previous year, while late-stage deals plummeted even further.

How to Evaluate Desktop AI Assistants for Your Workflow

• Data Privacy Model: Determine whether the assistant processes your information locally on your device or sends it to cloud servers. Local processing reduces exposure to data breaches and regulatory fines but may limit performance.
• Integration with Your Operating System: Check how seamlessly the assistant integrates with your existing desktop environment. Incumbents like Microsoft Copilot and Apple's Siri already have deep OS integration, which new entrants must overcome.
• Real-World Use Cases: Evaluate whether the assistant handles your actual daily tasks, from drafting emails to scheduling meetings, rather than just responding to generic commands.

The pullback in AI investment has forced startups to focus on tangible use cases rather than futuristic visions. Investors are no longer chasing moonshots; they are backing tools that solve real problems, even if those problems are mundane. This dynamic favors IrisGo's pragmatic approach.

The broader context is instructive. C3.ai Inc, a company that promised enterprise AI solutions, saw its market capitalization shrink from a 2021 peak of $15 billion to roughly $1.3 billion today. That cautionary tale underscores the risk of overpromising and underdelivering, a fate IrisGo must avoid by proving its desktop assistant can outperform both human labor and existing AI tools.

For IrisGo to succeed, it must overcome two formidable hurdles: user habit and competition. The desktop assistant market is fragmented, with incumbents such as Microsoft Copilot and Apple's Siri already embedded in operating systems. Convincing users to adopt a new assistant, even one that promises privacy and efficiency, will require more than a clever algorithm; it will demand seamless integration into daily routines.

Yet the competition is not static. Google and Microsoft are rapidly embedding AI into their desktop environments, while open-source alternatives like Ollama and LM Studio are lowering the barrier to entry for local AI. IrisGo's best chance may lie in the enterprise segment, where privacy concerns and compliance requirements create a natural demand for its model. A pilot with a Fortune 500 company, something Ng's network is well-positioned to facilitate, could provide the validation needed to scale.

Ng's recent public statements emphasize "efficient, accessible AI," a refrain that resonates with enterprises wary of cloud bills and regulators concerned about data sovereignty. By anchoring IrisGo in these megatrends, Ng is not merely investing in a startup; he is betting on a category. The challenge, however, will be execution: turning a technical proof-of-concept into a product that millions of users instinctively reach for daily.