IrisGo Builds an AI Desktop Buddy

IrisGo is trying to answer a question that Microsoft, Apple, Google, and OpenAI are all circling from different angles: after chatbots, what should an AI assistant actually become? The startup, backed by Andrew Ng’s AI Fund, has raised a $2.8 million seed round to build a desktop companion that learns how people work on their PCs and automates repeat tasks with minimal prompting.

The pitch is more ambitious than another chatbot docked to a screen. IrisGo says its software can watch a user’s workflow, understand context across apps, and turn repeated actions into autonomous “skills.” Its own product page describes the system as an AI coworker that learns workflows and runs them across screens, apps, and local systems.

The AI PC Needs a Real Use Case

IrisGo is arriving as the tech industry pushes the idea of the AI PC, where local chips such as NPUs and GPUs handle AI tasks directly on the device. That matters because PC makers need more than hardware benchmarks to convince users to upgrade. They need software that feels meaningfully different from cloud chatbots.

IrisGo’s answer is local workflow automation. AI Fund describes the company as an on-device assistant built for AI PCs and powered by local AI models. IrisGo also promotes its product as “AIPC-ready,” using local NPU and GPU compute for faster and more private inference.

A Small Startup in a Platform War

The challenge is that IrisGo is entering territory already being claimed by much larger companies. Microsoft has Copilot and Copilot+ PCs, Apple is integrating AI deeper into macOS and iOS, Google is extending Gemini across Workspace and Chrome, and OpenAI is increasingly becoming a general-purpose work assistant.

That makes IrisGo’s positioning both risky and interesting. Instead of trying to own one app, it wants to sit across the desktop experience. Dealroom reported that the company has secured Acer as its first OEM partner, with its software expected to be preloaded on devices, giving the startup a possible distribution path that most early AI companies do not have.

Privacy Will Decide User Trust

The product’s biggest selling point could also become its biggest concern. Any assistant that watches desktop activity, reads local context, or learns workflows has to convince users that it is useful without feeling invasive.

That is why IrisGo is emphasizing on-device processing. The company’s local-first message directly addresses one of the main anxieties around AI assistants: whether sensitive work data must leave the machine to become useful. If IrisGo can make local automation reliable, it may find a stronger opening in business workflows than consumer novelty.

Broader Impact

The larger significance is that AI software is moving from answering questions to performing tasks. The first wave of generative AI helped users write, summarize, code, and search. The next wave is trying to remove repetitive digital work altogether.

For IrisGo, the near-term test is simple: can it turn daily PC behavior into dependable automation without breaking trust? If it can, the company may become an early example of what AI PCs are actually for. If it cannot, the same idea will likely be absorbed into Windows, macOS, or productivity suites built by companies with deeper platform control.