Meet Iris: The AI Infrastructure Letting You Automate almost Anything
It all started with a simple idea: What if QA testing could just… handle itself?
We built Iris as an automated QA testing tool powered by Anthropic’s models. But something unexpected happened. Users kept asking, "Can this do more?" Turns out, the real magic wasn’t the QA automation—it was the underlying tech: AI agents that actually understand and interact with interfaces, like humans do.
So we pivoted. Hard.
Instead of focusing on one use case, we rebuilt Iris as infrastructure for AI agents. Now, it works with any model (not just Anthropic) and any app or device (thanks to Omniparser v2, our screen-parsing engine). Need to automate a finicky website, a mobile app, or even legacy software? Iris gets it done—no API access required.
Why this matters: Most "AI automation" tools are rigid and brittle. Iris is different. It’s designed for the messy, unpredictable real world. Whether you’re automating customer support, data entry, or workflows, you can use Iris.
Where we’re going:
"Desktop apps (starting with Mac Silicon!): Automate any task on your machine. *[P.S. Windows/Linux versions are already cooking-tell us your OS so we can prioritize!]*"
SaaS-ready infrastructure: Let companies build custom AI agents at scale.
Faster, cheaper, bulletproof: 100% of early revenue is being reinvested to cut costs and boost reliability.
This isn’t just about saving time. It’s about giving people superpowers.
We’re shipping updates fast-if you’ve ever thought, "I wish I could automate this… but it’s impossible," we want to hear from you. Let’s break some "impossible" together.
Bro that's awesome. I am not so formal in commenting so sorry in advance. I asked it make a list of all the places where i could buy best quality low cost woolen clothes in Bulk in Delhi and it crafted a beautiful list with info and contacts from Google.
Iris is super cool! Feels like having a smart assistant that helps you get things done on your computer. Big shoutout to the team! Excited to see where this goes! 🙌
This pivot’s gold—realizing the infrastructure was the product is a classic founder epiphany. How are you tackling edge case reliability, especially with inconsistent UI elements across platforms? That’s where most agents tap out.
Replies
OpenCopilot
Meet Iris: The AI Infrastructure Letting You Automate almost Anything
It all started with a simple idea: What if QA testing could just… handle itself?
We built Iris as an automated QA testing tool powered by Anthropic’s models. But something unexpected happened. Users kept asking, "Can this do more?" Turns out, the real magic wasn’t the QA automation—it was the underlying tech: AI agents that actually understand and interact with interfaces, like humans do.
So we pivoted. Hard.
Instead of focusing on one use case, we rebuilt Iris as infrastructure for AI agents. Now, it works with any model (not just Anthropic) and any app or device (thanks to Omniparser v2, our screen-parsing engine). Need to automate a finicky website, a mobile app, or even legacy software? Iris gets it done—no API access required.
Why this matters:
Most "AI automation" tools are rigid and brittle. Iris is different. It’s designed for the messy, unpredictable real world. Whether you’re automating customer support, data entry, or workflows, you can use Iris.
Where we’re going:
"Desktop apps (starting with Mac Silicon!): Automate any task on your machine. *[P.S. Windows/Linux versions are already cooking-tell us your OS so we can prioritize!]*"
SaaS-ready infrastructure: Let companies build custom AI agents at scale.
Faster, cheaper, bulletproof: 100% of early revenue is being reinvested to cut costs and boost reliability.
This isn’t just about saving time. It’s about giving people superpowers.
We’re shipping updates fast-if you’ve ever thought, "I wish I could automate this… but it’s impossible," we want to hear from you. Let’s break some "impossible" together.
Bro that's awesome. I am not so formal in commenting so sorry in advance. I asked it make a list of all the places where i could buy best quality low cost woolen clothes in Bulk in Delhi and it crafted a beautiful list with info and contacts from Google.
Great job by the developers and team !!!
EverTutor AI
this is called an gold great work team @shanur_rahman
Blink Eye
Can't wait to test Iris on Windows. Great Demo. Just craving me to test on Windows. Wish I had Mac to test instantly.
OpenCopilot
@nomandhoni use this image to run it locally on windows
https://hub.docker.com/r/shanurcsenitap/operator
Iris is super cool! Feels like having a smart assistant that helps you get things done on your computer. Big shoutout to the team!
Excited to see where this goes! 🙌
This is big! I can't stop thinking about the possibilities with this. This changes everything.
Don't think try it out
Woop woop! Can't wait to test it on CRM automation.
This pivot’s gold—realizing the infrastructure was the product is a classic founder epiphany. How are you tackling edge case reliability, especially with inconsistent UI elements across platforms? That’s where most agents tap out.