IrisGo is an early-stage startup with six employees, $2.8 million in the bank, and a simple idea: your computer should already know how to do the things you do every day, without you having to explain it each time.
The software sits on your Windows or Mac desktop and watches how you work. You demonstrate a task once, such as pulling data from one system and pasting it into a report, and IrisGo records the steps and stores them as a reusable workflow. From that point on, it can run the task on its own, triggered either on a schedule or when certain conditions are met. No coding, no setup forms, no instructions written in plain English to a chatbot.
This is a meaningfully different approach from today's popular AI tools. ChatGPT, Gemini, and Microsoft Copilot all require you to ask them something. You type a prompt, they respond, you take the output and do something with it. IrisGo sits in the background and acts. The closest competitors in that space are Anthropic's Claude Cowork and OpenAI's Operator, both of which can also control a computer and execute multi-step tasks. But those are cloud-first products built by well-funded giants. IrisGo's angle is that most of the processing stays on the device itself.
That privacy architecture is the detail most likely to matter to corporate buyers. When an AI tool runs in the cloud, your emails, invoices, internal documents, and workflow patterns travel to an external server to be processed. IrisGo processes most of that locally, on your machine. Cloud processing is used for more complex tasks, but only when the user explicitly approves it. For industries where data governance matters, such as insurance, legal, finance, or healthcare administration, that is a real distinction worth examining.
The market these companies are racing into is large and moving quickly. Gartner forecasts that AI-capable PCs will account for 55% of all global PC shipments by 2026, up from 31% in 2025. By 2029, they are expected to become the default. Gartner also projects that by end of 2026, 40% of software vendors will prioritize building tools that run directly on PCs, up from just 2% in 2024. IrisGo is positioning itself ahead of that shift.
The Acer pre-installation deal is the strategic move to watch. Getting software onto a device before the user sets it up is the fastest way to acquire users at scale without spending on marketing. If IrisGo can replicate that deal with two or three additional laptop makers, it gains a distribution advantage that no amount of advertising can easily match.
That said, the competitive pressure here is not theoretical. Microsoft's Copilot is built into Windows 11. Anthropic's Claude Cowork can read and write local files on both Mac and Windows. ChatGPT's Agent Mode can control a cloud-based virtual computer to execute tasks on your behalf. These are products backed by hundreds of billions in capital. IrisGo, with six people and a seed round, is asking to be the tool that learns your specific workflows at a company level, in a way that keeps your data private.
For business operators thinking about where this fits: the near-term question is not whether to use IrisGo specifically, but whether your team has started mapping which repetitive computer tasks could be handed off to software entirely. Email triage, report assembly, invoice routing, data entry between systems: these are the tasks this category of tool is designed to absorb. The tools are now good enough to take on real workflows. The organizations that identify those tasks now, and start testing tools that can handle them, will be six to twelve months ahead of those that wait for the technology to feel more settled.