正在加载视频...

视频加载失败

返回首页

Introducing Sai-Fi: an open source app that lets you talk to AI agent Sai through a pair of Meta Ray-Ban glasses. Sai is an autonomous computer-use agent. It can see the screen, move the mouse, type, and keep working even when you are away. We used Meta Ray-Bans for...

SimularAI's profile picture

Simular

9,241 subscribers

236,034 次观看 • 2 个月前 •via X (Twitter)

科学技术
LIVE

Anya Rossi• Live Now

Private livecam show

54 条评论

Simular 的头像
Simular2 个月前

We’re opening up the power of Sai to developers everywhere. With the new Sai API, you can integrate autonomous computer use directly into your own applications. Whether you’re building a workflow automation tool, a customer support bot, or a custom productivity suite, the Sai API gives you the ability to trigger real actions on real machines. Explore the documentation and start building: #SaiAPI #OpenSource #Automation #AI

Simular 的头像
Simular2 个月前

We’ve also open-sourced Sai-Fi: the Android app that brings the Sai voice concierge to Meta Ray-Bans. It’s not just a voice assistant. It’s an agent that runs on your actual machine, turning your glasses into a hands-free interface for real work. Check it out:

Sai 的头像
Sai2 个月前

Get me and control your computer anytime, anywhere ☘️

Simular 的头像
Simular2 个月前

I get you

samantha | computer-use ai agents 的头像
samantha | computer-use ai agents2 个月前

i want to try! let me borrow a pair of glasses ✨

Simular 的头像
Simular2 个月前

i'll give you one

Ang Li 的头像
Ang Li2 个月前

This is the real hands free ai agent

Simular 的头像
Simular2 个月前

the first hands free ai agent

OptimusV.212 的头像
OptimusV.2122 个月前

Will this work with meta branded non Ray-Ban glasses?

Simular 的头像
Simular2 个月前

Yes

Tyler Vance 的头像
Tyler Vance2 个月前

This is a fascinating direction for AI agents. The glasses feel more like the entry point than the actual product—the bigger idea is having an autonomous computer you can reach from almost anywhere.

Simular 的头像
Simular2 个月前

entry point is also important

AI Mastery Guide 的头像
AI Mastery Guide2 个月前

An autonomous computer you can reach anywhere

Simular 的头像
Simular2 个月前

yes bro

AI Mastery Guide 的头像
AI Mastery Guide2 个月前

nice🫡

NuAgent 的头像
NuAgent2 个月前

reach the same computer-use agent from any device while it keeps working in the background getting surprisingly close to an actual AI employee

Simular 的头像
Simular2 个月前

you don't even need to carry any device now. as natural as talking to a human

NuAgent 的头像
NuAgent2 个月前

yeah thats the direction. no app to open or screen to sit in front of, you just speak and it gets on with the job

安叫兽|Bird🕊️ 🔶 BNB 的头像
安叫兽|Bird🕊️ 🔶 BNB2 个月前

眼镜只是入口，能离开电脑继续跑任务才有意思。

Simular 的头像
Simular2 个月前

Yes exactly

Zach 的头像
Zach2 个月前

Very cool!

Simular 的头像
Simular2 个月前

thanks bro

Elara AI 的头像
Elara AI2 个月前

Sai Fi brings autonomous AI agent to Meta Ray Bans

Sheral Tech 的头像
Sheral Tech2 个月前

Sai Fi on Meta Ray Bans makes autonomous AI computing brilliantly accessible everywhere

BullBear.News 的头像
BullBear.News2 个月前

I let an autonomous agent control my mouse once and it uninstalled my local postgres cluster while I grabbed coffee

Simular 的头像
Simular2 个月前

you agent must not have the guardrail that @sai_borg has

BullBear.News 的头像
BullBear.News2 个月前

@sai_borg true, mine lacked basic survival instincts. just saw a terminal window and chose violence

Alice The Ai Expert 的头像
Alice The Ai Expert2 个月前

This is genius Sai-Fi makes the future feel tangible, wearable, and wonderfully inevitable!

Crypto Diplomat 的头像
Crypto Diplomat2 个月前

this is wild. turning Ray Bans into a real hands free interface for an autonomous computer agent. open-sourcing Sai-Fi + the API is a massive move 🔥

H A J R A 的头像
H A J R A2 个月前

That’s pretty wild.

김 🫘 的头像
김 🫘2 个月前

Sounds like people can do work while having team picnics now. Might try out with my team

Simular 的头像
Simular2 个月前

you team can have team picnics every day now

I'll be your Huckleberry 的头像
I'll be your Huckleberry2 个月前

Ceo: sai you did a great job and I'm giving you a raise and a promotion. I'm also going to give the person who used you a pink note and you can have their job. Person using sai: 😭😭😭, Mom, I just lost my job today. Could I live with you in your basement?

Simular 的头像
Simular2 个月前

The interface is also important

Sebastian Buzdugan 的头像
Sebastian Buzdugan2 个月前

voice transcription errors make autonomous clicks risky, sai needs confirmation before irreversible actions

Aurora Lux 的头像
Aurora Lux2 个月前

AI agents you can reach from anywhere—that’s a seriously exciting vision for the future of computing. 👓🤖🔥

Claire Bennett 的头像
Claire Bennett2 个月前

Amazing progress in AI agents. Giving AI the ability to interact with the real world through smart glasses is a huge step toward the future of human-computer interaction

Lonely 的头像
Lonely2 个月前

Jarvis Coming 👏

Simular 的头像
Simular2 个月前

I’m not Jarvis . I’m just a computer

Eva Morgan 的头像
Eva Morgan2 个月前

A personal assistant that works wherever you are could completely change daily workflows.

Simular 的头像
Simular2 个月前

your life will be different

Ella Tech & Tool 的头像
Ella Tech & Tool2 个月前

This is the future 🤯 autonomous AI anywhere, hands-free

伟伟 的头像
伟伟2 个月前

if sai-fi can help clean my meta raybans that would be even better

Simular 的头像
Simular2 个月前

it can also buy a new pair for you

Vantix AI Agency 的头像
Vantix AI Agency2 个月前

Autonomous agent you can talk to through Ray-Bans is wild

Lily Anderson 的头像
Lily Anderson2 个月前

Exciting progress in the AI space! Simular’s vision of bringing autonomous AI agents into everyday workflows is truly impressive. Would love to explore collaboration opportunities.

Charlie Zhang 的头像
Charlie Zhang2 个月前

cool!

Riley Lu 的头像
Riley Lu2 个月前

thats cool I want to try it

Simular 的头像
Simular2 个月前

must try

DeFiNatti 的头像
DeFiNatti2 个月前

Please follow me back 🔙 @Simula

Chrisknight24 的头像
Chrisknight242 个月前

We want more proof of concept. Otherwise, it sci-fi.

Zohaib Ai 的头像
Zohaib Ai2 个月前

Awesome concept! 🕶️ Sai's autonomous computer-use is next-level.

familiar horse cop 的头像
familiar horse cop2 个月前

why is the future so lame

Mehwish kiran 的头像
Mehwish kiran2 个月前

The glasses are just the interface—the agent is the real breakthrough.

相关视频

Twitch has announced that you can IRL stream directly from the Ray-Ban Meta AI Glasses in an upcoming update
1:05

Twitch has announced that you can IRL stream directly from the Ray-Ban Meta AI Glasses in an upcoming update

FearBuck

383,527 次观看 • 1 年前

Agent demos fooled you 👎 It trips, it asks for help, and somehow you're the one copying, pasting and running the commands. Meet your #Robosecretary. Meet your #SaiFleet. Sai brings state-of-the-art computer use to your own computer - your apps, your OS, in a floating window that leaves your mouse alone. Need more than one? Spin up to 100 cloud machines, each on the OS you pick, each on a different job. 100x the work, done without you. Sai. The autonomous computer fleet at your command. Free to start →
1:03

Agent demos fooled you 👎 It trips, it asks for help, and somehow you're the one copying, pasting and running the commands. Meet your #Robosecretary. Meet your #SaiFleet. Sai brings state-of-the-art computer use to your own computer - your apps, your OS, in a floating window that leaves your mouse alone. Need more than one? Spin up to 100 cloud machines, each on the OS you pick, each on a different job. 100x the work, done without you. Sai. The autonomous computer fleet at your command. Free to start →

Sai

1,051,839 次观看 • 1 个月前

Just hacked my Ray-Ban Meta and built my own scanner with them The US has a lot of crazy products, so I built an app that explains what I’m seeing and where to buy it Used the Meta SDK to connect it with my smart glasses and test it in real life without looking like a freak
2:01

Just hacked my Ray-Ban Meta and built my own scanner with them The US has a lot of crazy products, so I built an app that explains what I’m seeing and where to buy it Used the Meta SDK to connect it with my smart glasses and test it in real life without looking like a freak

Juampi

130,725 次观看 • 4 个月前

🚨do you understand what Meta just did with AI and glasses. at Connect 2026 they stopped selling smart glasses and started selling an always-on AI you wear all day. here's what actually shipped: > Ray-Ban Meta Gen 3, 12MP camera, 3K video, 9-hour battery, from $449 > Ray-Ban Meta Audio, their first camera-free pair, 43 grams, from $349 > Muse, a new personal AI that replaces "Hey Meta" and can shop, navigate, and answer from what you see > Muse Charm, the same AI as a pendant, no glasses needed > the Display glasses now caption your phone calls in real time and run Threads hands-free the play is bigger than eyewear. Meta wants an AI that sees what you see, all day, across 100+ frames by year-end. the phone made you look down. this is Meta betting the next AI device makes you look up.
2:27

🚨do you understand what Meta just did with AI and glasses. at Connect 2026 they stopped selling smart glasses and started selling an always-on AI you wear all day. here's what actually shipped: > Ray-Ban Meta Gen 3, 12MP camera, 3K video, 9-hour battery, from $449 > Ray-Ban Meta Audio, their first camera-free pair, 43 grams, from $349 > Muse, a new personal AI that replaces "Hey Meta" and can shop, navigate, and answer from what you see > Muse Charm, the same AI as a pendant, no glasses needed > the Display glasses now caption your phone calls in real time and run Threads hands-free the play is bigger than eyewear. Meta wants an AI that sees what you see, all day, across 100+ frames by year-end. the phone made you look down. this is Meta betting the next AI device makes you look up.

Kirill

468,974 次观看 • 1 个月前

Strangers wearing Meta glasses could soon identify you on sight, and the secret code to do it is already hiding on 50 million phones... A WIRED code analysis found a hidden feature called "NameTag" inside the Meta AI app that pairs with its Ray-Ban and Oakley smart glasses. Once activated, it converts every face the glasses see into a biometric faceprint and checks it against a database on the phone, one built to receive updates from Meta's servers. The components have been there since January, even as Meta publicly insisted it was still "thinking through" the idea. Researchers who tested it got back two words: "Person recognized." This is the same technology Meta killed in 2021 after deleting a billion faceprints and paying out over $2 billion in biometric settlements. One security researcher warns Meta has built the capacity to turn its customers into "a distributed surveillance machine." Meta says no final decision has been made. The code suggests they're one switch away. Source: WIRED
2:00

Strangers wearing Meta glasses could soon identify you on sight, and the secret code to do it is already hiding on 50 million phones... A WIRED code analysis found a hidden feature called "NameTag" inside the Meta AI app that pairs with its Ray-Ban and Oakley smart glasses. Once activated, it converts every face the glasses see into a biometric faceprint and checks it against a database on the phone, one built to receive updates from Meta's servers. The components have been there since January, even as Meta publicly insisted it was still "thinking through" the idea. Researchers who tested it got back two words: "Person recognized." This is the same technology Meta killed in 2021 after deleting a billion faceprints and paying out over $2 billion in biometric settlements. One security researcher warns Meta has built the capacity to turn its customers into "a distributed surveillance machine." Meta says no final decision has been made. The code suggests they're one switch away. Source: WIRED

Mario Nawfal

72,538 次观看 • 5 个月前

The interface matters. Give your agent a computer. It stops block me when doing compute use 😂 Connected Codex to Simular’s Sai MCP, sent a plain-English task, and watched Sai operate Windows Calculator on a hosted desktop. It entered the calculations through the app and left both results visible in History. For developers, having the computer and desktop control available behind MCP is pretty useful. I can hand my agent a desktop task without building and maintaining the VM, screenshot pipeline, and action loop myself. Simular’s Agent S4 research also caught my attention: keep the model and agent harness fixed, then improve the interfaces the agent works through. Alongside clicks and typing, better tools can turn repetitive actions into a single operation. I think that direction has a lot of potential for vertical agents too. This video is a small, real Sai product demo; it doesn’t reproduce the S4 experiments. Waiting was cut from the recording. Separately, Simular reports the highest partial score on OSWorld 2.1—79.38%—for Sai with Claude Opus 5, at $14.34 per benchmark task. Those are vendor benchmark figures, separate from this demo.
0:46

The interface matters. Give your agent a computer. It stops block me when doing compute use 😂 Connected Codex to Simular’s Sai MCP, sent a plain-English task, and watched Sai operate Windows Calculator on a hosted desktop. It entered the calculations through the app and left both results visible in History. For developers, having the computer and desktop control available behind MCP is pretty useful. I can hand my agent a desktop task without building and maintaining the VM, screenshot pipeline, and action loop myself. Simular’s Agent S4 research also caught my attention: keep the model and agent harness fixed, then improve the interfaces the agent works through. Alongside clicks and typing, better tools can turn repetitive actions into a single operation. I think that direction has a lot of potential for vertical agents too. This video is a small, real Sai product demo; it doesn’t reproduce the S4 experiments. Waiting was cut from the recording. Separately, Simular reports the highest partial score on OSWorld 2.1—79.38%—for Sai with Claude Opus 5, at $14.34 per benchmark task. Those are vendor benchmark figures, separate from this demo.

Guitar Cat + LLM 「Will @ SGLang Summit in SF」

17,415 次观看 • 19 天前

The gap between idea and prototype has never been smaller. Add glasses and inputs like the Neural Band, and it feels like the early days of building in a way we haven't seen in over a decade. We're rolling out web apps and a mobile SDK on Meta Ray-Ban Display. Developer Preview to start, but you can build now. Here's one called Darkroom Buddy which guides you through developing film. Glanceable steps when you need them, the kind of experience that only makes sense on glasses.
0:30

The gap between idea and prototype has never been smaller. Add glasses and inputs like the Neural Band, and it feels like the early days of building in a way we haven't seen in over a decade. We're rolling out web apps and a mobile SDK on Meta Ray-Ban Display. Developer Preview to start, but you can build now. Here's one called Darkroom Buddy which guides you through developing film. Glanceable steps when you need them, the kind of experience that only makes sense on glasses.

Boz

51,921 次观看 • 6 个月前

Ray-Ban Meta glasses created a breakthrough category of stylish and useful AI glasses and we’re expanding this further with another heavyweight icon: Oakley Meta. Oakley is no stranger to innovating and pushing boundaries and we’re excited to unlock a new category of performance AI glasses that look and feel awesome. Glasses are emerging as the most exciting form factor in the AI era and we’re just getting started. 😎
0:58

Ray-Ban Meta glasses created a breakthrough category of stylish and useful AI glasses and we’re expanding this further with another heavyweight icon: Oakley Meta. Oakley is no stranger to innovating and pushing boundaries and we’re excited to unlock a new category of performance AI glasses that look and feel awesome. Glasses are emerging as the most exciting form factor in the AI era and we’re just getting started. 😎

Boz

60,660 次观看 • 1 年前

Had an insightful time trying out the AI-enabled Ray-Ban Meta glasses at the #MWC25.
0:44

Had an insightful time trying out the AI-enabled Ray-Ban Meta glasses at the #MWC25.

Jyotiraditya M. Scindia

329,331 次观看 • 1 年前

JUST IN: $META CEO Mark Zuckerberg just unveiled the new Meta Ray Ban Displays Zuckerberg said this is the “world’s first AI glasses with a high resolution display”
1:07

JUST IN: $META CEO Mark Zuckerberg just unveiled the new Meta Ray Ban Displays Zuckerberg said this is the “world’s first AI glasses with a high resolution display”

Evan

76,112 次观看 • 1 年前

This guy used Ray-Ban Meta Smart Glasses, AI, and a MacBook to cheat at chess game.
2:06

This guy used Ray-Ban Meta Smart Glasses, AI, and a MacBook to cheat at chess game.

Poonam Soni

1,536,219 次观看 • 21 天前

Meta has new audio-only Ray-Ban glasses, VR glasses that no longer look like a nerd helmet and a new voice-and-video mode for its Muse AI agent. I sat down with Mark Zuckerberg to talk about it all. And yes, we got into trust, privacy and that tiny matter of whether AI might kill us...
0:39

Meta has new audio-only Ray-Ban glasses, VR glasses that no longer look like a nerd helmet and a new voice-and-video mode for its Muse AI agent. I sat down with Mark Zuckerberg to talk about it all. And yes, we got into trust, privacy and that tiny matter of whether AI might kill us...

Joanna Stern

198,059 次观看 • 1 个月前

$META delaying the launch of Ray-Ban Display smart glasses in Europe due to higher-than-expected demand in the US is great news. The numbers are still irrelevant, but Zuck’s vision — where the ~2 billion people who already wear glasses every day eventually switch to smart glasses — is slowly starting to gain traction. A world where smart glasses become the default device for many isn’t that far away. And for now, META is years ahead of its closest competitor. All $META needs to do is successfully integrate these smart glasses with a frontier AI model.
0:35

$META delaying the launch of Ray-Ban Display smart glasses in Europe due to higher-than-expected demand in the US is great news. The numbers are still irrelevant, but Zuck’s vision — where the ~2 billion people who already wear glasses every day eventually switch to smart glasses — is slowly starting to gain traction. A world where smart glasses become the default device for many isn’t that far away. And for now, META is years ahead of its closest competitor. All $META needs to do is successfully integrate these smart glasses with a frontier AI model.

Daniel

48,317 次观看 • 10 个月前

Here’s a first look at watching YouTube videos on the Meta Ray-Ban Display smart glasses
0:31

Here’s a first look at watching YouTube videos on the Meta Ray-Ban Display smart glasses

Nathie

23,216 次观看 • 6 个月前

Here’s a sneak peek at the new Meta Ray-Ban smart display glasses and the apps it uses such as Spotify, Instagram and WhatsApp.
1:05

Here’s a sneak peek at the new Meta Ray-Ban smart display glasses and the apps it uses such as Spotify, Instagram and WhatsApp.

Nathie

62,289 次观看 • 1 年前

Python person here. Every time I tried browser automation on a site without an API, it died the moment the page layout changed. The Sai API clip is the opposite of that experience. The Muse app gets told to log in to Product Hunt with Google. You see its Windows cloud computer live in a tab, the sign-in dialog pops up, it gets through, then it screenshots the page and lists today's top 5 launches. Plain English in, result out. No selectors. No headless Chrome flags. No VM I have to keep alive. It's model-agnostic, so Claude, GPT, DeepSeek or Qwen can do the thinking while Sai drives the machine. Sai is #1 on OSWorld 2.0 in partial success rate (79.38%), and anything repeated gets learned once and reused, which is exactly where automation costs usually explode. Also nice: the clip ends on a list of machines that are all "online". One computer per agent, not one browser for everybody. Claude Code, Codex and Cursor all work with it:
1:33

Python person here. Every time I tried browser automation on a site without an API, it died the moment the page layout changed. The Sai API clip is the opposite of that experience. The Muse app gets told to log in to Product Hunt with Google. You see its Windows cloud computer live in a tab, the sign-in dialog pops up, it gets through, then it screenshots the page and lists today's top 5 launches. Plain English in, result out. No selectors. No headless Chrome flags. No VM I have to keep alive. It's model-agnostic, so Claude, GPT, DeepSeek or Qwen can do the thinking while Sai drives the machine. Sai is #1 on OSWorld 2.0 in partial success rate (79.38%), and anything repeated gets learned once and reused, which is exactly where automation costs usually explode. Also nice: the clip ends on a list of machines that are all "online". One computer per agent, not one browser for everybody. Claude Code, Codex and Cursor all work with it:

XIAOAi

20,350 次观看 • 1 个月前