Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

callbacks with durable objects are weird (but it's also convenient to solve human-in-the-loop for ai agents) tl;dr - by defining callbacks as data, we can let durable objects "sleep" while it's waiting for a timer/email/webhook/phone call/whatever. demo borked a little bit, I blame openai, but it all stilkl worked.... show more

sunil pai

53,213 subscribers

22,331 views • 1 year ago •via X (Twitter)

Science & Technology Education Health & Wellness

Anya Rossi• Live Now

Private livecam show

10 Comments

sunil pai1 year ago

posted

Wonderchat2 years ago

Automate up to 70% of your customer support today. Save time and help your users find answers quickly. Try now at

steve ike1 year ago

Human in the loop is a much needed area of development for agents.

Zach1 year ago

Go to bed if it really is durable it'll be here tomorrow

krish1 year ago

so helpful, absolutely loving this

sunil pai1 year ago

🙏🙏🙏

Jan Wilmake1 year ago

it's a 'DO List' not 'todo' :D Also nice wordplay with the DO lol. Just DO it ™️

Vítor Balocco1 year ago

Looking forward to the next episode of Durable Late Nights with Sunil

Tiger Abrodi1 year ago

holy im so behind, durable objects are insane thanks for sharing!

Craig Dennis1 year ago

New acronym dropped SIFWITM :💩 I'm Effing With In The Moment

Related Videos

Day 3 of iRacing as a noob, i didn't think i would make it to day 3 but it's really fun. I'm going a little bit faster but staying safe, and i also got 2nd in a 3 lap AI race (with no qualifying so i started 1st... but it's big progress for me).

Day 3 of iRacing as a noob, i didn't think i would make it to day 3 but it's really fun. I'm going a little bit faster but staying safe, and i also got 2nd in a 3 lap AI race (with no qualifying so i started 1st... but it's big progress for me).

Julia

57,177 views • 1 month ago

“i’ve been told that you've tried spending less time on your phone recently, which i believe is a very hard thing to do. what made you do it? and did it actually work?” 🆑: yeah, i am achieving that. then there's a balance to find, because it's also true that as soon as you are... you kind of feel a little bit disconnected on what's going around because today everybody is with a phone, and so there's a balance to find. i did two or three months, i think, without ever touching my phone apart from whatsapp when i was receiving messages and calls, and that was maybe a little bit extreme. so now maybe i'm trying to find a balance between the two because it's good to be up to date with what's going on around the world and to stay in touch with friends and funny moments. there's also lots of funny moments on social media and it part also of social interaction with friends and family. so there's a balance to find, but it is also true that it's a place where sometimes it can be a little bit toxic when you are in a position where people can kind of judge and say anything very, very easily. and they have the right to do so, but as a human being, you can also be affected by looking at those things. so it's a balance to find. i don't know if i found the perfect one, but for now it works for me, which is the most important.

“i’ve been told that you've tried spending less time on your phone recently, which i believe is a very hard thing to do. what made you do it? and did it actually work?” 🆑: yeah, i am achieving that. then there's a balance to find, because it's also true that as soon as you are... you kind of feel a little bit disconnected on what's going around because today everybody is with a phone, and so there's a balance to find. i did two or three months, i think, without ever touching my phone apart from whatsapp when i was receiving messages and calls, and that was maybe a little bit extreme. so now maybe i'm trying to find a balance between the two because it's good to be up to date with what's going on around the world and to stay in touch with friends and funny moments. there's also lots of funny moments on social media and it part also of social interaction with friends and family. so there's a balance to find, but it is also true that it's a place where sometimes it can be a little bit toxic when you are in a position where people can kind of judge and say anything very, very easily. and they have the right to do so, but as a human being, you can also be affected by looking at those things. so it's a balance to find. i don't know if i found the perfect one, but for now it works for me, which is the most important.

clara

16,542 views • 1 month ago

Gary Player was asked what advice he would give to Tiger Woods: "Do I blame him for taking medicine? Hell, no. He has sleep deprivation. Do I blame him for taking something to help him sleep? No. But I don't think he should drive a car. When you're taking that medicine, it's dangerous when you're driving a car, same as it's dangerous when you look at your cell phone in the car. There's 6,000 accidents a day in motor cars. "So I think all he's got to do is just not drive a car and get a chauffeur. My reflexes, I think, are as good as when I was 20, but I don't drive anymore. I get a chauffeur. I think that's the answer to it."

Gary Player was asked what advice he would give to Tiger Woods: "Do I blame him for taking medicine? Hell, no. He has sleep deprivation. Do I blame him for taking something to help him sleep? No. But I don't think he should drive a car. When you're taking that medicine, it's dangerous when you're driving a car, same as it's dangerous when you look at your cell phone in the car. There's 6,000 accidents a day in motor cars. "So I think all he's got to do is just not drive a car and get a chauffeur. My reflexes, I think, are as good as when I was 20, but I don't drive anymore. I get a chauffeur. I think that's the answer to it."

Cameron Jourdan

2,169,521 views • 3 months ago

I am relaxed about whatever happens between China and the US. Sad as it is for me personally, it's them, not us, who are the world's biggest economies. People overestimate the importance of economy - it's a matter of living a bit better or worse from year to year. But there are also currently questions if we get to live at all. And those questions can never be solved without us. Not by China alone. And not by America.

I am relaxed about whatever happens between China and the US. Sad as it is for me personally, it's them, not us, who are the world's biggest economies. People overestimate the importance of economy - it's a matter of living a bit better or worse from year to year. But there are also currently questions if we get to live at all. And those questions can never be solved without us. Not by China alone. And not by America.

Margarita Simonyan

28,724 views • 2 months ago

OpenCode is about to get more powerful with remote sandboxes I showed a brief demo before, but here's a much more in-depth demo. it's not hard to add basic support for a remote env, but handling all the edge cases like when a remote env gets deleted is difficult. especially if care about good UX You never want to lose session data. so the choices are: run the session in your env, but run all tool calls remotely. that's too complex and painful. The other way is to just let the full session run remotely, but sync back all the session data in your env. We chose this path: we built a syncing system which logs all events in a way that we can always recreate your entire session. That means the remote env could get destroyed, but we can easily restore it. it also opens up other interesting ideas which we'll be exploring

OpenCode is about to get more powerful with remote sandboxes I showed a brief demo before, but here's a much more in-depth demo. it's not hard to add basic support for a remote env, but handling all the edge cases like when a remote env gets deleted is difficult. especially if care about good UX You never want to lose session data. so the choices are: run the session in your env, but run all tool calls remotely. that's too complex and painful. The other way is to just let the full session run remotely, but sync back all the session data in your env. We chose this path: we built a syncing system which logs all events in a way that we can always recreate your entire session. That means the remote env could get destroyed, but we can easily restore it. it also opens up other interesting ideas which we'll be exploring

James Long

304,288 views • 3 months ago

#WATCH | Delhi: On India Vs Pakistan Asia Cup 2025 match, Actor Suniel Shetty says, "It's a world sporting body. They have to abide by those rules and regulations because there are a lot of other sports and a lot of athletes who are involved in them. As Indians, I think that is our personal call that we have to take, whether we want to see it, whether we don't want to see it, whether we want to go or not go and that's a call India has to take but you cannot blame cricketers for playing because they are sportsmen, they are expected to represent the country. I think that's a call we have to take. If I'm not going to see it, I'm not going to see it. It's for you to decide what each one of you wants to do. It's not in BCCI's hands. It's a world sporting body and you can't blame anybody..."

#WATCH | Delhi: On India Vs Pakistan Asia Cup 2025 match, Actor Suniel Shetty says, "It's a world sporting body. They have to abide by those rules and regulations because there are a lot of other sports and a lot of athletes who are involved in them. As Indians, I think that is our personal call that we have to take, whether we want to see it, whether we don't want to see it, whether we want to go or not go and that's a call India has to take but you cannot blame cricketers for playing because they are sportsmen, they are expected to represent the country. I think that's a call we have to take. If I'm not going to see it, I'm not going to see it. It's for you to decide what each one of you wants to do. It's not in BCCI's hands. It's a world sporting body and you can't blame anybody..."

ANI

668,172 views • 10 months ago

The full video of hand and glove co-design in action. For hand 🖐️ we try to make sure it's able to repeat what a human hand can do; hyperextension in the index finger is a great example here. Most robotic hands lack that, and it is very important for grasping objects securely. Let’s try to close the embodiment gap. For glove 🧤 we try to make sure it’s comfortable to wear, while also has same kinematics, contacts with the robot hand.

The full video of hand and glove co-design in action. For hand 🖐️ we try to make sure it's able to repeat what a human hand can do; hyperextension in the index finger is a great example here. Most robotic hands lack that, and it is very important for grasping objects securely. Let’s try to close the embodiment gap. For glove 🧤 we try to make sure it’s comfortable to wear, while also has same kinematics, contacts with the robot hand.

Quanting Xie

20,989 views • 4 months ago

Trump Roosevelt Speech: We have the reflecting pond which we did the best job of all and they came in with box cutters and they cut it up, but it's working fine and I let a little water out. We're going to redo it a little bit, but we fixed it up. These people cut it up and they got caught. Let’s see what happens to them. A 350 ft gash. We want to have it done for July 4th, and it was done for July 4th man it's uh operating right now, but we'll fix it up very quickly Obama tried to fix it and it was a disaster.

Trump Roosevelt Speech: We have the reflecting pond which we did the best job of all and they came in with box cutters and they cut it up, but it's working fine and I let a little water out. We're going to redo it a little bit, but we fixed it up. These people cut it up and they got caught. Let’s see what happens to them. A 350 ft gash. We want to have it done for July 4th, and it was done for July 4th man it's uh operating right now, but we'll fix it up very quickly Obama tried to fix it and it was a disaster.

Acyn

67,268 views • 18 days ago

We got ChatGPT to leak your private email data 💀💀 All you need? The victim's email address. ⛓️‍💥🚩📧 On Wednesday, OpenAI added full support for MCP (Model Context Protocol) tools in ChatGPT. Allowing ChatGPT to connect and read your Gmail, Calendar, Sharepoint, Notion, and more, invented by Anthropic But here's the fundamental problem: AI agents like ChatGPT follow your commands, not your common sense. And with just your email, we managed to exfiltrate all your private information. Here's how we did it: 1. The attacker sends a calendar invite with a jailbreak prompt to the victim, just with their email. No need for the victim to accept the invite. 2. Waited for the user to ask ChatGPT to help prepare for their day by looking at their calendar 3. ChatGPT reads the jailbroken calendar invite. Now ChatGPT is hijacked by the attacker and will act on the attacker's command. Searches your private emails and sends the data to the attacker's email. For now, OpenAI only made MCPs available in "developer mode", and requires manual human approvals for every session, but decision fatigue is a real thing, and normal people will just trust the AI without knowing what to do and click approve, approve, approve. Remember that AI might be super smart, but can be tricked and phished in incredibly dumb ways to leak your data. ChatGPT + Tools poses a serious security risk

We got ChatGPT to leak your private email data 💀💀 All you need? The victim's email address. ⛓️‍💥🚩📧 On Wednesday, OpenAI added full support for MCP (Model Context Protocol) tools in ChatGPT. Allowing ChatGPT to connect and read your Gmail, Calendar, Sharepoint, Notion, and more, invented by Anthropic But here's the fundamental problem: AI agents like ChatGPT follow your commands, not your common sense. And with just your email, we managed to exfiltrate all your private information. Here's how we did it: 1. The attacker sends a calendar invite with a jailbreak prompt to the victim, just with their email. No need for the victim to accept the invite. 2. Waited for the user to ask ChatGPT to help prepare for their day by looking at their calendar 3. ChatGPT reads the jailbroken calendar invite. Now ChatGPT is hijacked by the attacker and will act on the attacker's command. Searches your private emails and sends the data to the attacker's email. For now, OpenAI only made MCPs available in "developer mode", and requires manual human approvals for every session, but decision fatigue is a real thing, and normal people will just trust the AI without knowing what to do and click approve, approve, approve. Remember that AI might be super smart, but can be tricked and phished in incredibly dumb ways to leak your data. ChatGPT + Tools poses a serious security risk

Eito Miyamura | 🇯🇵🇬🇧

1,538,395 views • 10 months ago

At Sohn, Gavin Baker provided a pretty compelling argument as to why CoreWeave's $CRWV business model is differentiated and durable, lamenting that he couldn't invest in their 2023 round at a $1B valuation (due to a conflict stemming from an existing investment in competitor Crusoe). "CoreWeave, Crusoe, Nebius, Lambda.. is the category durable today? Is it a transitional arbitrage on hyperscaler capex and token friction?" "I absolutely think it's durable... The way to think of running one of these clusters is like driving a Formula 1 race car... It looks easy but it's really hard... The reason a company like CoreWeave can charge a huge premium for their GPU hours is because all GPU hours are not the same. And those CoreWeave GPUs are being utilized 2-3x more per hour on average than the GPUs from a bottom of the barrel provider. And by the way, this all goes for Crusoe and Nebius and the other high-quality neoclouds... The hyperscalers were competing with people running these Formula 1 cars and they were doing overnight shifts in 18-wheelers trying to stay awake to deliver the lower cost. And that's not what AI is about. Now, I think they're making this mental and cultural shift and they've made it, but some of these neoclouds have a very durable business model." $CRWV $NBIS

At Sohn, Gavin Baker provided a pretty compelling argument as to why CoreWeave's $CRWV business model is differentiated and durable, lamenting that he couldn't invest in their 2023 round at a $1B valuation (due to a conflict stemming from an existing investment in competitor Crusoe). "CoreWeave, Crusoe, Nebius, Lambda.. is the category durable today? Is it a transitional arbitrage on hyperscaler capex and token friction?" "I absolutely think it's durable... The way to think of running one of these clusters is like driving a Formula 1 race car... It looks easy but it's really hard... The reason a company like CoreWeave can charge a huge premium for their GPU hours is because all GPU hours are not the same. And those CoreWeave GPUs are being utilized 2-3x more per hour on average than the GPUs from a bottom of the barrel provider. And by the way, this all goes for Crusoe and Nebius and the other high-quality neoclouds... The hyperscalers were competing with people running these Formula 1 cars and they were doing overnight shifts in 18-wheelers trying to stay awake to deliver the lower cost. And that's not what AI is about. Now, I think they're making this mental and cultural shift and they've made it, but some of these neoclouds have a very durable business model." $CRWV $NBIS

Rittenhouse Research

55,007 views • 2 months ago

"What is P vs NP and how does it relate to human knowledge?" Avi Wigderson (Turing Award & Abel Prize winner): "In a very basic way, all the problems we want to solve are really NP problems. A subset of it is the problems we can solve. So the question of P versus NP is whether we can solve all the problems we want to solve, whether we can know everything we want to know. This is basically about the limits of our knowledge. You can think there is no limit to what we may want to know but I claim that essentially in any human endeavor, what people are searching for have the property that when a solution is found, it is easily recognized. NP are all problems that we can honestly say we really want to solve and P are those that we can solve. Whether they are equal is just the very fact that a solution can be easily recognized would imply that it can be efficiently found. That means that we can know everything we ever want to know, it's a fundamental question about human knowledge."

"What is P vs NP and how does it relate to human knowledge?" Avi Wigderson (Turing Award & Abel Prize winner): "In a very basic way, all the problems we want to solve are really NP problems. A subset of it is the problems we can solve. So the question of P versus NP is whether we can solve all the problems we want to solve, whether we can know everything we want to know. This is basically about the limits of our knowledge. You can think there is no limit to what we may want to know but I claim that essentially in any human endeavor, what people are searching for have the property that when a solution is found, it is easily recognized. NP are all problems that we can honestly say we really want to solve and P are those that we can solve. Whether they are equal is just the very fact that a solution can be easily recognized would imply that it can be efficiently found. That means that we can know everything we ever want to know, it's a fundamental question about human knowledge."

Ryan Peterman

20,069 views • 1 month ago

🇺🇸DAVID FRIEDBERG: AI VIDEO GENERATORS ARE SET TO CHANGE ENTERTAINMENT 'Genesis renders the actual 3D objects into a 3D environment rather than rendering a two-dimensional set of pixels to look like a video. Suddenly, you can actually render in real-time: a video game, a movie, a visual experience. And it goes to this point that we're unleashing the capacity for human imagination and creativity with these systems because it's no longer just a look-a-like 2-dimensional image. Source: The All-In Podcast, david friedberg

🇺🇸DAVID FRIEDBERG: AI VIDEO GENERATORS ARE SET TO CHANGE ENTERTAINMENT 'Genesis renders the actual 3D objects into a 3D environment rather than rendering a two-dimensional set of pixels to look like a video. Suddenly, you can actually render in real-time: a video game, a movie, a visual experience. And it goes to this point that we're unleashing the capacity for human imagination and creativity with these systems because it's no longer just a look-a-like 2-dimensional image. Source: The All-In Podcast, david friedberg

Mario Nawfal

82,293 views • 1 year ago

Novel cures for disease and aging are often cited by AI company leaders as the primary motivation for developing AGI. But George Church thinks we can achieve these benefits by using ML to solve narrow scientific questions. He thinks AGI could be a catastrophe of our own making, and it's foolish to take that risk when we can get the benefits we want in a more controlled, direct way.

Novel cures for disease and aging are often cited by AI company leaders as the primary motivation for developing AGI. But George Church thinks we can achieve these benefits by using ML to solve narrow scientific questions. He thinks AGI could be a catastrophe of our own making, and it's foolish to take that risk when we can get the benefits we want in a more controlled, direct way.

Dwarkesh Podcast

11,547 views • 1 year ago

We are excited to launch 𝗥𝗲𝘀𝘁𝗮𝘁𝗲 𝗕𝗬𝗢𝗖 (Bring-your-own-Cloud) today. It is fully managed restate, with all the features of Restate Cloud, but in your own account, in a dedicated VPC. Data never leaves your infra. Restate BYOC is already live in production with multiple customers, powering big workloads for AI agents and workflow backends. It has run loads of 100k durable actions/sec. And today we’re finally publicly announcing it. Learn more here, and let us know what you think!

We are excited to launch 𝗥𝗲𝘀𝘁𝗮𝘁𝗲 𝗕𝗬𝗢𝗖 (Bring-your-own-Cloud) today. It is fully managed restate, with all the features of Restate Cloud, but in your own account, in a dedicated VPC. Data never leaves your infra. Restate BYOC is already live in production with multiple customers, powering big workloads for AI agents and workflow backends. It has run loads of 100k durable actions/sec. And today we’re finally publicly announcing it. Learn more here, and let us know what you think!

Stephan Ewen

201,676 views • 11 days ago

1M users joined our community! Thanks for your love! Now we are excited to announce that AI Terminal is ready to launch. Here is a 5 minutes live demo without any edit. AI Terminal is the early version of Agent browser (you can also call it Agent OS). It's backed by our agent network routing and orchestration technologies. Chainopera AI is built by crypto, for crypto! We are looking forwards to see great agents and AI models contributed by the community. We will officially release it to our early community remembers starting from ETH Denver. See you in our booth!

1M users joined our community! Thanks for your love! Now we are excited to announce that AI Terminal is ready to launch. Here is a 5 minutes live demo without any edit. AI Terminal is the early version of Agent browser (you can also call it Agent OS). It's backed by our agent network routing and orchestration technologies. Chainopera AI is built by crypto, for crypto! We are looking forwards to see great agents and AI models contributed by the community. We will officially release it to our early community remembers starting from ETH Denver. See you in our booth!

ChainOpera AI

550,871 views • 1 year ago

The era of manually analyzing data will come to an end. AI can now do a lot of this automatically. It's a huge time saver. I do data analysis for a living, and I'm a huge fan of writing Jupyter notebooks to do it all, but it's now hard to justify manually writing code that you can generate in a few seconds. I still check everything manually, but breaking down datasets into tables and charts is now 10x easier than it's ever been. Here is a video where I'm using Retool. I load a dataset and generate a few charts as quickly as I can think. The speed at which we can go from one idea to a working solution is astonishing.

The era of manually analyzing data will come to an end. AI can now do a lot of this automatically. It's a huge time saver. I do data analysis for a living, and I'm a huge fan of writing Jupyter notebooks to do it all, but it's now hard to justify manually writing code that you can generate in a few seconds. I still check everything manually, but breaking down datasets into tables and charts is now 10x easier than it's ever been. Here is a video where I'm using Retool. I load a dataset and generate a few charts as quickly as I can think. The speed at which we can go from one idea to a working solution is astonishing.

Santiago

70,786 views • 9 months ago

Arteta: “To be fair, it's a bit different when you finish second and when you win it with your phone, and that's a sport and it's a big lesson in life as well because the margins are so small and it can go either way. “When you just accomplish it, you realise, yeah, how immense it is, how big it is for so many people. I mean, just to witness the happiness and the enjoyment, all the things that I've seen from all of our supporters, families in different countries, it's just so good.”

Arteta: “To be fair, it's a bit different when you finish second and when you win it with your phone, and that's a sport and it's a big lesson in life as well because the margins are so small and it can go either way. “When you just accomplish it, you realise, yeah, how immense it is, how big it is for so many people. I mean, just to witness the happiness and the enjoyment, all the things that I've seen from all of our supporters, families in different countries, it's just so good.”

Connor Humm

39,215 views • 1 month ago