IrisGo Is the AI That Learns Your Job by Watching

IrisGo, a new "AI operating system" that learns your workflow by watching you do it, just took the number one Product of the Day spot on Product Hunt, and it is aimed squarely at solopreneurs drowning in admin. The startup launched IrisGo for Solopreneurs on October 7, 2026, with a simple promise: demonstrate a task once, and the software handles it forever. There are no repeated prompts, and your sensitive data stays on your device, according to the company's Product Hunt launch post.

The timing is hard to miss. A generation juggling side hustles, freelance clients, and full-time jobs now has an AI tool built specifically for the unglamorous middle of the workday: inboxes, invoices, meeting notes, and expense reports. IrisGo's bet is that the next useful assistant will not live in a chat window waiting for instructions. It will sit closer to the operating system, learn routines by observation, and run them in the background while you do the work that actually matters.

What IrisGo actually does

The product ships as a desktop app with five headline features. A Daily Brief pulls top priorities from email, calendar, and to-do lists. Email Triage sorts what needs an urgent reply from what can wait. A continuously updated Radar keeps every to-do in one list. Meeting Summary turns calls into notes with action items attached. And Watch & Learn, the centerpiece, converts a demonstrated routine into a reusable workflow that runs again with new inputs.

"Our Watch & Learn feature is central to understanding how you work, and handling that task forever," the company wrote in its launch post. On the launch thread, the makers answered detailed questions about how the replay handles real-world messiness: unlike a rigid macro, it treats a demonstration as a pattern, recognizes the list or folder being worked from, and processes items one by one until none remain. One commenter asked about downloading candidate resumes, renaming them, and sorting them into folders; another wanted lecture transcripts turned into study guides. The team said both are in scope.

The software is free during the public beta and runs on Windows and Apple Silicon Macs. A Linux version is on the longer-term roadmap, but the team says it is stabilizing the two main platforms first, as reported by OpenAIMaster.

Why this launch is getting attention

IrisGo was co-founded by Jeffrey Lai, a former Apple engineer who helped develop the Chinese-language version of Siri. Even the company name is a reversal of Siri, as reported by IMP News. The startup raised 2.8 million dollars in seed funding led by AI Fund, the venture studio founded by Andrew Ng, to build software that reduces repetitive office work through AI automation.

That pedigree explains the Product Hunt reception. The launch took the number one Product of the Day spot on October 7, as reported by OpenAIMaster, with commenters calling out exactly the problem IrisGo targets: small, specific tasks too niche for a dedicated integration but costly when repeated by hand. For young workers, the appeal is concrete. Workers say they want AI their jobs will not allow, and much of that frustration sits in the gap between a chatbot that answers questions and a tool that actually does the clicking.

There is also a philosophical shift worth noting. Most AI tools still ask users to translate messy office work into prompts, then paste the output somewhere else. IrisGo moves toward a different model, in which the assistant learns from patterns: drafting an email from a document, pulling figures into a report, or repeating a routine that usually takes five tabs and too many clicks, as reported by Startup Fortune.

The catch: it watches you work

The same feature that makes IrisGo interesting is the one that should give pause. Software that learns by observing everything on your screen is, by definition, software with deep access to your work life. The company says workflows and sensitive data stay on the device, but earlier reporting noted a hybrid architecture, with cloud processing only when explicitly authorized and end-to-end encrypted, as reported by OpenAIMaster.

That tension is the real story. As Startup Fortune put it, the hard part of this category is not making a model sound helpful; it is getting software to move reliably through real user interfaces without creating privacy, security, or trust problems. An assistant with permission to watch your screen, read your inbox, and click through your apps is powerful precisely because it is trusted, and trust has to be earned feature by feature.

For now, the practical calculus is simple. If you are a freelancer or side-hustler spending hours each week on inbox triage, meeting summaries, and expense filing, a free beta that promises to learn those chores by watching is worth a look. Go in clear-eyed: this is a public beta, so expect rough edges, and on-device processing means your machine does the heavy lifting. The broader question — how much of your workday you want an AI to see — is one every new tool in this category will have to answer. Find more ideas for working smarter on our productivity page.