Sai is the world's first robosecretary that works with a fleet of autonomous computers, doing your endless screen work. Each computer navigates software the way you do: reads the screen, clicks the button, types in the form. Trained to interact with the computer interface, Sai automates software even when they lack APIs: legacy desktop apps, internal portals, anything behind a login. Sai supports Windows, macOS and Linux. 73% on OSWorld.
Sai is the world's first robosecretary that comes together with a fleet of autonomous computers doing your endless screen work. We believe the future of work isn't just one computer — it's many computer.
Each computer navigates software the way you do: reads the screen, clicks the button, types in the form. That's the whole point. Because Sai works at the interface layer, it automates things that have no API and never will — legacy desktop apps, internal portals, anything behind a login. Windows, macOS, and Linux. 73% on OSWorld.
What that buys you in practice:
🖥️ Real computers, not a chat box. You hand over a task, it runs it on an actual desktop. Open the workspace any time to watch, or take the mouse back mid-run.
🔁 It learns the route once. Any finished task can be saved as a skill, scheduled, or triggered by an event — so the tenth run is faster and cheaper than the first.
🔒 Autonomy with a seatbelt. Tiered approvals, "trust for this task," and encrypted input for passwords and codes — the model never sees plaintext.
The agent we actually optimize for is whether it's still useful and affordable the tenth time you ask.
Tell us the screen work that eats your week — we'll tell you honestly whether Sai can take it off your hands.
a fleet of computers doing the boring work is the right framing 🔥 how does it handle flaky legacy UIs?
This is very very cool. As a CEO, this is something I'd be interested to use, but how does it build trust?
I like that it remembers how I like things done instead of making me re-explain every time, which saves token usage a lot.
Having followed Simular's previous releases, Sai feels like a massive step forward for agentic workflows. What immediately stands out is the practical focus on operating at the interface layer—being able to handle legacy desktop apps, internal portals, and tasks behind logins without needing custom APIs is a huge win.
A few highlights that really make this release stand out:
True Parallel Fleet Architecture: Running tasks across multiple dedicated cloud computers rather than a single local instance makes scaling complex workflows seamless.
Smart Snapshot & Cost Efficiency: Reducing token usage without sacrificing screen-reading accuracy brings the cost per task down significantly (and makes the free tier viable).
High OSWorld Benchmark Score: Scoring 73% on OSWorld 2.0 while staying lower in cost per task than competing models shows serious engineering optimization behind the scenes.
Thoughtful Safety & Adaptation: The ability to gracefully pause, ask for human input on major UI redesigns, and use tiered approvals builds real trust for production use.
Excited to see how the team continues to refine handling edge cases and dynamic layouts. Congrats to the whole Simular team on another great launch! 🚀
Hey PH — I'm Richard (Chih-Lun), part of the crew helping ship Sai.
It's been fascinating to watch the evolution: local machines, then a shared VM, then dedicated VMs for each of us, and now a full agent fleet for everyone.
Huge thanks to everyone who helped get this out the door. Really grateful to have been part of it.
A final file is useful, but I also need to know how it was made. Can I see which apps Sai opened and which files it changed? Is there a run log I can share with a teammate?
Congrats on the launch, Simular team!
How do you guys ensure that all runs are deterministic while still having this "thinking" part when, let's say, the form changes and we have to enter another value? Thinking from a perspective where I have to fill out hundreds of website forms almost daily, but those forms are updated monthly.
@Simular@agidude Congrats one the PH launch ! , Does it have human in loop control / fully autonomous ?
I've been waiting for something like this that doesn't need a ton of setup just to automate basic desktop tasks, hope it delivers
I’m Yulia, a designer at Simular. One thing I kept coming back to while designing Sai: when the computer starts doing the work for you, the interface has to change too. You’re no longer designing screens for someone to operate - you’re designing a way for people to delegate, understand what’s happening, and stay in control.
That shift has been one of the most interesting design problems I’ve worked on. Sai can operate computers on its own, but making that autonomy feel understandable, trustworthy, and genuinely useful is what turns the technology into a product.
Very excited that Sai is finally out in the wild! Try it, break it, and tell us what feels weird. I’ll be in the comments all day! 🚀
Our internal tool barely has an API, so most automation products stop before the real useful part.
A computer that can work through the app itself opens up tasks we have left manual for years.
My work rarely stays in one tab. I go from email to a native app to a spreadsheet just to finish one request. If Sai is able to work across the desktop, then it will be very useful to me.
Congrats on the launch, @justin2025 ! If Sai gets better at my usual file names and checks over time, that could save a surprising amount of setup.
Do you have any becnchmarks for accuracy on common tasks like booking or form filling?
Do you see this replacing RPA tools or is it a different category entirely?
About Sai on Product Hunt
“The autonomous computer fleet at your command”
Sai launched on Product Hunt on September 21st, 2026 and earned 417 upvotes and 48 comments, earning #3 Product of the Day. Sai is the world's first robosecretary that works with a fleet of autonomous computers, doing your endless screen work. Each computer navigates software the way you do: reads the screen, clicks the button, types in the form. Trained to interact with the computer interface, Sai automates software even when they lack APIs: legacy desktop apps, internal portals, anything behind a login. Sai supports Windows, macOS and Linux. 73% on OSWorld.
Sai was featured in Productivity (663.6k followers) and Artificial Intelligence (481.6k followers) on Product Hunt. Together, these topics include over 301.4k products, making this a competitive space to launch in.
Who hunted Sai?
Sai was hunted by Justin Jincaid. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how Sai stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
Justin Jincaid
Hey Product Hunt 👋
I'm Ang and I build Sai at Simular.
Sai is the world's first robosecretary that comes together with a fleet of autonomous computers doing your endless screen work. We believe the future of work isn't just one computer — it's many computer.
Each computer navigates software the way you do: reads the screen, clicks the button, types in the form. That's the whole point. Because Sai works at the interface layer, it automates things that have no API and never will — legacy desktop apps, internal portals, anything behind a login. Windows, macOS, and Linux. 73% on OSWorld.
What that buys you in practice:
🖥️ Real computers, not a chat box. You hand over a task, it runs it on an actual desktop. Open the workspace any time to watch, or take the mouse back mid-run.
🔁 It learns the route once. Any finished task can be saved as a skill, scheduled, or triggered by an event — so the tenth run is faster and cheaper than the first.
🔒 Autonomy with a seatbelt. Tiered approvals, "trust for this task," and encrypted input for passwords and codes — the model never sees plaintext.
The agent we actually optimize for is whether it's still useful and affordable the tenth time you ask.
Tell us the screen work that eats your week — we'll tell you honestly whether Sai can take it off your hands.