Cloudflare Just Shipped 20+ Features for AI Agents in One Week

Hello everyone and welcome to this week In net is the special April 2026 agents week edition. We have a special episode about our week full of announcements. One of our innovation weeks. It's the first of its kind. So the first agents week. I'm your host Ronto based in Lisbon, Portugal. And with me I have again for the second time in the show Ming and Annie. Hello. Welcome. Hello. Hello again. So excited to be back. Hello again. And I bet you're really tired because innovation weeks are all about deadlines and shipping and sharing with the world what we do, but also really busy. So, how busy and full are you? Uh, I feel we I feel like last week was was really the the road show moment. you know, we did a lot of kind of talking about all the cool stuff that was coming down coming out um with a bunch of internal teams, some external folks, and then this week, I think, at least for me, was more about um making sure that the thing that we were shipping got over the line uh because one of the project one one of the announcements was was related to AI gateway. I think this week Yeah. I think this week was what was really interesting as well. Like I feel like we still had like announcements trying to like being slotted in as this week was coming along and so we were like trying to figure out like the best day to position them. But like once we kind of have the general story like set in stone and figure out when to go out, a lot of our blog authors and like the teams that work on it, they're super responsible and super timely. So we were able to get everything in on time. So no no no no worries there. That's good. and last minute blogs, last minute things that we ship, those are always the case for innovation weeks. And we also have we're recording on a Friday and this episode will come out on Monday or Tuesday. On Monday, we'll also have a few other posts, right? Yes. Yeah. Um yeah, we're know we have so many things and so many people, you know, wanted to share what they're building that it had a we had it spill over and extend it a bit to Monday. We might have something for Tuesday as well. It will be extended. It's not the first time. Uh sometimes that happens although for a while now I we we have kept it from Monday to Friday in the past and birthday weeks are an example of that we go over because there's so many things and agents developers there's there's a lot there so it makes perfect sense. Last week we spoke about how the week is all about Cloudflare being the best place to build deploy secure agents. Uh so we're actually bringing uh the full platform infrastructure not only to developers but also to agents and that helps the developers work but also everyone working with agents these days we're all developers even non-developers in a sense. What can we say in terms of now that the blog posts are out what is the main takeaways that people should take from the this full week who wants to start? I can start. Yeah. I I think the the the main takeaway or at least kind of how I see it is that um that the that that as you're building agents, you know, agents really need a different kind of infrastructure to run it. They're fundamentally just very different kinds of software than kind of like the cloud, the 1.0 version of the cloud was designed to serve. And in this new world, you're going to need need new primitives and just kind of new tooling to support this like very very new use case and this this thing happening at like such a larger scale. And uh I think the week was was yeah largely about this is what Cloudflare is providing for this new era of of software. Makes sense. Any addition Annie? Yeah, I think one of the my favorite parts about the welcome post that we that that Rita and Dane had put out was like they did like a quick map. what is the scale of where we expect you know agent infrastructure to need and it's you know if we have x amount of knowledge workers around the US each of them are running x amount of agents then it comes down to around you know needing million of compute instances to run these agents and what's really interesting is like in the previous model kind of serves many users but when you have an agent that needs its own compute instance to be able to run itself and also run the code that it's running that scale becomes much greater. How do we do that sustainably? And I feel with the platform we've built on workers, it's u unknowingly we kind of built the next generation of infrastructure. Exactly. Actually, I was hearing today from Kenton Varda on Twitter. He was me me me me me me me me me me me me me me ment mentioning how and he was the creator at start of workers nine years ago and he was actually mentioning how he created containers, sandboxes work things like that that at the time were not u completely relevant for this age agents because this age wasn't here but now those are actually really perfect for this. So it's interesting how things that were done a while ago actually are now actually perfect for for this age. This reminds me and this in this podcast podcast we had many shows many episodes including with John Graham coming which I started this podcast with talking about how the internet wasn't built initially for security for a number of things that were added. Now we're in the agents era in a sense and it was not all also built for agents initially but with the layers some already there others are just improving it could actually be better in terms of efficiency now because there will be more internet traffic in general because agents will be performing more and more things but also in many other things it's uh it's really more relevant than ever some tools that were around and new tools that are coming. So quite interesting to see the the weak forming that perspective in in particular and the blog post from Retita and Dane you you mentioned defines that specifically for sure. One of the things actually that the week starts it's about cloth's agentic cloud right mostly about from developers to coding agents now and the importance of of the agentic cloud what can we say about that first day that Monday also with cool announcements with our uh Monday announcements I think the big story we wanted to tell around it was really focus on kind of like the compute primitive um and all the different compute primitives you could use to kind um build your agents with. We also slipped in something really interesting there uh around uh CLI for all of Cloudflare. Um so I guess like the the the one that I would would really want to highlight is um well our our sandboxes, our containers um they went G and we actually have two announcements around that. So one is like yeah we have way more capabilities built into the sandbox XDK now where it you can very much almost give like your agent a a uh a full computer right so if you it needs a file system it needs like to run terminal commands it can really easily do that uh with sandboxes and I think what's really interesting um kind of going back to the story uh you you told around like the first generation of internet wasn't really built with like security mindset I would say like that we're actually kind of we we're seeing that with agents as well like um because like you know the the industry is moving so fast um people are starting to realize oh shoot we really need to kind of like lock this down have a way to monitor um all the like security um elements like what what can an agent have access to and so on so forth. We've added um you know outbound worker as like an egress proxy to our sandboxes so that you can get that like monitoring for what your agents uh are are trying to uh get access to outside of the sandbox. There's many cool blogs uh this week for sure uh and this day for for sure and what those are may make make people be able to do agents be able to do for people in the sense uh it's it's quite a astonishing for those who don't know for example how important it is building a CLI for all cler for example this has some repercussion online what can we explain folks about it. Yeah. Yeah. I saw so much excitement around um you know uh people hearing that we'll not have a CLI for all of Cloudflare. So for those who who didn't know um like uh for for for a while I feel our CLI experience has been a little disjointed in the sense that you know for the developer platform we had Wrangler and Wrangler only contained like uh uh APIs for you know the developer platform but you know Cloudflare is so much more. We we are we also have like application security performance products as well. And so with the Cloudflare CLI, this is a very early technical preview of what we're trying to build. Um we'll have CLI experience for every single product on Cloudflare. And so um what makes this super exciting is that you know agents um almost like know how to run you know commands really easily. A and so when you're able to give kind of like this one simple interface to interact with uh Cloudflare I uh agents will be really easily uh be able to use it to you know spin up workers you know set up a domain for it um be able to you know put put CDN in front of it all all with uh Cloudflare CLI. It's it's quite impressive. Uh for those that are less technical, the the thing that I find more interesting about this is this is technical, but what this uh enables is not that technical. is actually making things really easy and fast and you having the ability or agent for you actually having the ability uh to go over the products and the API operations and there's so many there with an ease that is uh takes a lot of from you of you needing to understand and try to do specific things. It's uh quite interesting to see all about also the single code mode MCTP server uh specifically here uh and Cloudflare's entire API is really big. So this is makes a difference, right? Yeah, exactly. Um and I think uh as we continue to work on it, we'll we'll add more to it as well. So if you have any feedback for the early technical preview, let us know on Discord, let us know on socials and uh we'll be sure to incorporate it. Uh should we move on to um Tuesday? What can we say there? Um so Tuesday very much focused on um kind of the now the security story. Um uh like I kind of mentioned, you know, um especially if you're trying to use uh agents on like an enterprise skill, you really need to understand like what it has access to. So for example, if you want to have an agent that you know writes code for you or employs resources for you, you might be giving it access to like internal databases in infrastructure to lock that down. And how do you make sure it's doing it in a way that you you want it to? And so I think we have a couple of really interesting announcements here. The big call out for the day I would say is Cloudflare Mesh. Um this one is I think the one that uh a lot of people might have heard already. Um it's uh uh our own like private networking service that lets you connect you know your agents uh your devices um all together in a private network and so any sort of access between these two um nodes would be happening within a private secured environment. Uh those are also relevant for sure. um in in in terms of the CFL mesh uh it's it's also aggregating right uh in in terms of workflows and terms of abilities right you can connect like uh various like uh services uh to it and it can all talk to each other within this mesh network so um uh you you know you can you can connect uh services from um you know private VPCs you have on AWS you can connect it uh connect it to services within like you know the Cloudflare developer platform. Uh you can connect your own personal devices and then they can all talk to each other. Um I'm not sure if that answered your question. Sure. Sure. Uh but I also seen a bit of repercussion there in terms of caller mesh in action. Uh people like connecting a virtual machine, a laptop and an agents SDK project running on Clare workers to the same private network for secure access. for example, there's there's a bunch of examples that many folks from Calfare and others have been sharing uh on Twitter that is quite interesting to see as well uh the ability of of what is put together in a sense, right? Yeah. I like the way that I I think is like really interesting to understand from maybe more of like a a individual like personal scale is like um a lot of people running like OpenClaw um a few months ago um you know in order to do that securely have your open cloud be able to talk to you let's say like your mobile device mesh would be like the product you would want um to have to kind of like like uh have that running in like a secure environment. Exactly. and people can get started. Usually all of the these posts has have a call to action at the end for people just to dip right in and and try it out in a sense, right? Anything we should mention more about Tuesday? I think another uh announcement I personally find super interesting is um the one that we have around scaling MCP adoption. Um in this blog post we share um how we personally run uh MCPs in uh in our own like you know Cloudflare as a company we have over like 4,000 employees and each of them may need uh access to you know uh you know the git gitlab uh when they're using a coding agent like cloud code or open code and so we share a reference architecture for how we set all of that up and how it gives us like the ability to monitor and control like what MCP servers um our our employees have access to and make sure like unauthorized servers don't get used. That's interesting as well. Actually, this week I spoke with Sharon Goldberg because we had a World Quantum episode on Tuesday uh because it was World Quantum Day. Uh and I I spoke with Sharon Goldberg that wrote this blog post uh about this specific topic uh in a very short way but uh also a cool perspective. What should we highlight more? If you scroll up, let me see. I think there was also a really interesting one on Tuesday around manage OOTH access uh make internal apps agent ready in one click. Um so that one uh was really interesting um because it it if you're familiar with kind of like Cloudflare access it allows you to um enable like kind of like a gateway in front of your applications that you want to be secured. don't want anyone from the internet to have access to. Um and like you know uh with access your users can enter like a password and username in order to authenticate that they should really have access to this uh application. Well now you know what's really interesting is um you know it's not just humans accessing apps and the web anymore. It's also agents. And so when agents um encounter, you know, an access wall such as the one we have at Cloudflare, typically they don't know what to do, right? They don't know your password. They don't know your they don't have your identity. They can't get into this app. And so what we highlighted here is essentially this new um you know managed OAS system. uh and it follows I believe like a a standard protocol which will give agents the ability to also um get access to these internal applications if you permit it to. Um and this blog outlines exactly how that happens and how you can give um your agent the the credentials it needs to access these apps. Yeah, it will save a lot of time do doing that specifically. Uh but you need to trust it of course. Um what more can we say potentially maybe going over to uh Wednesday agents need all of these things in order to bea in order to have the capacity to do like real tasks and that's what um like Wednesday and Thursday are all about uh I think so I guess for Wednesday looking at some of um some of the highlights I think the like one of the big announcements was project think which is kind of cloudflare's like next evolution for our agents SDK and I think there's a couple of really interesting parts to that. Um I think one is uh giving uh giving basically giving a framework for building longunning agents. Um most things that people are doing with agents like if you wanted to to have to handle a real world workflow uh is not something that takes like a you know a couple milliseconds and then you're finished. It's things that can take minutes or potentially even hours. And so, uh, this is a way for you to build kind of, um, agents with really durable execution where if it, um, suddenly gets shut down, it can kind of pick up where it left off. The other, I think, really interesting, um, uh, benefit or kind of like angle for, uh, project think is this like tiered execution environment concept, and that's all more about security. uh basically the ability to give agents the right tools with the like appropriate amount of permissions to do the thing that they need to. So um you know for example if you're doing something that only requires uh you to if you're processing something kind of locally you're munching some data you know you you you might want to just run code but you don't actually need to access the internet and whereas if you're kind of doing something that requires the whole operating system and requires access to the internet you might want to give it a sandbox. And so having these different kind of execution environments available means that there's always the like appropriately secured um component for the for the task. Um so yeah, I think that's something that like people were really excited about for uh for for for Wednesday and it has the safe sandboxes there helping keeping things protected as you said which is also cool. Um so yeah that that those that one also had a great repercussion online. There's excitement there for sure. Yeah. Yeah, definitely. I think like Yeah. Some of the other cool things for for Wednesday is our our improvements to browser run. So I think previously named browser rendering, we're giving a new name browser run. But basically, you know, I think agents also need access to a browser to be able to like fully realize all of their capabilities just because there are certain things you can only do in the browser. And what we've done with the improvements to browser run is add a lot more observability features. So as a person um you know running the agent you can we've added live view so you can kind of see what the agent is doing in the browser and just make debugging like a lot more easier. Recorded sessions um things like that. And then also uh there's now the ability for the agent to kind of hand off certain tasks to a human and then kind of then go back to its flow once the human's completed its step. Um so for example if the agent hits a capture that needs um human intervention we now have the mechanisms mechanisms to do that. Um we've also released within uh the the these new set of features like integrations with um uh the chrome death tools protocol. So now uh like different clients um like agents like you know cla uh claude and cursor and stuff can use can kind of have access to these APIs um and you know better control the control the browser through these things. And so yeah, I think like it's basically we've made browser run something that like agents can use really effectively. We've also increased the limits on the number of concurrent browsers you could have. Um kind of again think back to like this new scale that agents have or will need rather to uh in in in this new world. Really interesting uh interesting also this part of human in the loop perspective where the human can step in and interact with a page directly to click or type or navigate. uh even this week I think it was open AI that launched a bit of a of a product that uh allows for a its agent in a sense to be performing and moving around apps and and people can interact with that as well. So it's definitely an area that will potentially put computers more to use uh where you can have your agents do something and you also do something as well in a sense, right? Yeah. Yeah. Um maybe the last thing on Wednesday I just want to highlight is uh agent Lee um which is Cloudflare's agent that lives within the dashboard. Um and I think this is exciting for a couple of reasons. Um one because agent Lee is the is an agent that we've built on top of all of our primitives or all of our building blocks that we're also offering to customers. So I think it's like a cool way for people to see and kind of just be in get inspiration of what is possible when you're building on Cloudflare. Um and uh and and so you know for agently we use a lot of the things that we've talked about previously in the blog uh code mode with MCPs uh it it uses a lot of the models that we offer through workers AI. Yeah, we use our like MCP servers. Um it's built on like the agents SDK and durable objects and we share some of that uh like the architecture um and how we've built it in the blog post. So for other people who are looking to build maybe similar types of agents. Um, and then the other kind of really cool thing is it's just like a really useful tool for customers of Cloudflare. You know, I think particularly for people who are using a bunch of different cloud services, like if you have an app that you know, you've got a worker for and you've got you're using um uh R2 and all these things like when something goes wrong or if you're trying to diagnose something or just like find out information, you can use Agent Lee and Agent Lee can, you know, use the MCP and like use our tools to figure answer your questions rather than you having to click around in the dashboard yourself. Um that's super cool. the generative UI stuff is is like awesome to see. Basically building your own like it's kind of the beginning of building customized UI for users um for your you know your specific use case and task. Uh so yeah I think Asly is something that um I'm excited for it to like roll out to to more accounts for actually there's a bit of some examples here with natural language in terms of how you can use it and one of the uh interesting facts is that agent Lee has that name because it's named after Lee Holloway one of the founders of Gtherler. Um there's a wired story that people can find just search Lee Holloway and Wired and you'll see that story. It's quite a very incredible uh story for many reasons. So, you can learn about Lee Holloway a bit and why we're actually naming it agent Lee specifically uh with that story. Um well uh what where should we go next? One actually close to heart to be honest. It's the email service because I'm I've started using it for a few weeks now internally for a project called Randomness Playground uh all about entropy and walls of entropy and laval lamps and it's works really well. Uh it works really well. Yeah. Yeah. So um uh we can move on to some of the the stuff um that coming on Thursday. Yeah. Email service. I think we saw so much chatter on on on Twitter and and stuff like that about it. I think like so many people I think we're waiting for the open beta of our email service. So I think that was really exciting for a lot of the folks. Um, I mean it makes total sense like email itself as a protocol has been around longer than the web has and it's just kind of like this, you know, way like this kind of like default way to communicate between different like applications, right? People like a a notification is is is kind is you can kind of think of it as it is kind of like an API call. Email in a way, right? Structured email is kind of like an API call. Um, and yeah, now uh we're the the email service is in public beta. I think the really interesting like agents angle to this is we've also it has an integration with agents SDK um and uh agents SDK has an on email hook so you can have um basically an agent assigned to a email inbox like you know have all all your email sent to a certain alias go to an agent for to pro for processing. So, for example, if you set up a support company.com email address and you have uh you can really easily then funnel all of those emails that go to there to your support agent to then take actions on. Um, and so yeah, that's uh and I think we've also open sourced um kind of like an email client that we've built uh to kind of show how this might work uh for people who are also looking to build aic workflows that like are email are email based. True. And actually I saw some folks on Twitter saying, "Oh, this is I didn't know that Cloudflare actually had an email service before email routing and uh it's one of those services that's has many utilities, but this is now the full picture with email sending. You you'll have the the full picture and it's fully uh open and open source and free actually. You have a lot of you can do with a free version for it. Uh always great. And of course the the deploy to Cloudflare button for people to try it out. Uh and it to be honest it's also close to heart because the team that build it is here in Lisbon. So I've seen them talk about it many times which is cool. Where should we go next? Yeah I I think the other really really big story on Thursday that people were really excited about was artifacts. um which is our uh it's it's a our new storage primitive um basically version story storage that is like git compatible. Um yeah, I think this is I think also saw a lot of um excitement on on social media and stuff like that for for artifacts. This is basically a new storage primitive that we've released um for for agents and it is uh it's a it's like a version file system that speaks git and it's really built for agent scale. So I think a lot of you know version control systems today are kind of built for humans using it. humans who are maybe, you know, writing a couple of PRs a day, a dozen commits. Um, whereas agents can potentially, you know, commit code thousands of times within an hour. Um, and so that's that's what Artifax is for. It's it's a way to store the the code that an that an that an agent might generate. But not just code like anything that the agent is generating where you might want version control where you might want to be able to kind of like look at the different changes that the agent has make made be able to go back in time um and start from like a previous version, be able to like give uh give it to someone else and have them fork it. Um and so yeah, Artifacts is you know all about like doing this at scale um for for agents. It's definitely ready and and great for what agents can do. Uh very well prepared there. There's a few examples here uh about even how much code that will be generated because of of this new age. Um and uh that will have consequences and this will help deal with the possibilities here. Uh so really cool one for for sure and on hacker news it was email service and also artifacts and also another one callflare's AI platform an inference layer designed for agents those were really hot with people commenting and and really hot on hacker news. So yeah yeah the um AI platform um was was one of the ones that uh um that that my uh my team and myself were were working working on it. You know, I think whenever something goes on hacker news, I'm always like a little scared to read the comments cuz you you never know what what their take is going to be. Um, so it's good to see some positive positive comments about that. But yeah, uh the um we've uh we're basically kind of like you know reframing AI gateway um as this like in unified inference layer. And so the the big kind of announcements here are um a huge expansion to our model catalog um to a bunch of different model providers uh people like OpenAI um uh open open AAI and then also a lot of image model providers um like Miniax uh bite dance um also some like speechto text models uh and so now you know it's not just LLM that you can use through AI gateway but like all sorts of different kinds of models So you can really build these like multimodal applications um and you can you know get build for all of them through cloudflare and uh just you know it's the developer experience um with the workers mining is just really really simple if you want to switch models it's just like changing one line of code. So yeah this one is you know I'm obviously biased uh but it's it's very near near and dear to my to my heart. Ming, I I saw you you posted after that uh you already like do we already have like uh cloud opus 4 point? Mhm. Yeah. Yeah. Uh yeah. So yeah, also you know because you know we've got all of these models um as like the model catalog is not static obviously and so you know our goal now is like to have the you know the day that these new models drop have it available to run on Cloudflare um for for for our customers. and and it's great uh we have it internally as well and it's always great to have the new models there and I I like this little feature which is with AI gateway you'll get one centralized place a moder and manage AI spend always important to to see how much you're spending uh and the ability to change models or bring your model is quite important I I would say because there's model models that are more expensive uh you do a few tasks with them you can do another task with other models that are less expensive also important in this day and age I would say and there's actually a uh caller TV segment all all about this with Craig and Yuming uh that people can also check for sure yeah if you don't want to read the blog if video is more your thing we got you covered exactly where should we go next um uh yeah I think I think th those are the the like highlights for Thursday so we can like go to go to Friday which is the stuff that just came out this morning. Um I think yeah, one of the really ann exciting announcements for Friday is flagship um which uh is is Cloudflare's new uh new product that flag Cloudflare is launching which is our feature flagging system. Um I I think like the the the cool kind of like use case with flagship is you know if you think about like agents are writing so much code now um agents are uh you know people people are essentially just like having agents like build build their entire application and this kind of really makes the loop of like agents writing a code them deploying it to production and kind of like testing it you know in the browser and seeing what works and like looking at the logs makes it like even easier to do and or or rather like safer to do. So you can imagine agent writing a bunch of code then uh feature flagging just themselves or or you know like just themselves into that new thing that they've just built and then you know be able to then like test it within a browser and like kind of like click around. Um it just like really that can really speed up this development like loop that the agent has but all while you know not affecting your you know maybe main uh main main cohort of customers. um and kind of still keeping them on the like stable version. Uh I think that's like a really interesting use case for for this like feature flagging system. And one of the things I find really interesting is that many of the things that we spoke about before uh that are Clawflare related uh which is we're building many of these things with Cloudflare tools uh are what making what is making some of these tools also really relevant. For example, in this case, we we're using to to not break things uh things that Cloudflare does specifically. So, it's building CLER with Clare with a lot a lot of advantages in terms of uh durable objects. KV. So, there there's a bunch of of uh things of work building CULER with Clare that actually makes perfect sense for products like the these. I saw someone actually saying online uh that uh it's it's in this case it's your AI agent can write your code in minutes but someone still has to review merge deploy monitor and the agent can do that for you and there there this method allows you to move fast and if you want to see you can see and and uh review yourself but try not breaking stuff always important there as well right? Yeah. Yeah. And like kind of on your your point of like you know using Cloudflare to build Cloudflare. I think um I think like part of the reason why we're able to to to to ship so many things at Cloudflare and move so quickly is because we have really really strong primitives that we're building on like durable objects and workers and and stuff like that. And so we're able to ship things like this new feature flagging service like really quickly. Um but then also you know the things that we build a lot of it is born out of a need ourselves right. And so like you know we obviously internally we have a need to do feature flagging and you know we also want to be able to run this like nice little agent like uh software development loop as well and so you know that is a lot of times we're like trying to like solve our own problems and Friday is all about empowering the agentic web in a sense uh there's also a few other announcements uh that are great we all this is one that we publish every developer week and every innovation week which is network performance update trying to share how improved the network. But there's other less frequent uh announcements. I really enjoy this one which is introducing the agent readiness score. There's actually a website for that. You can actually put your website into this website and it will give you how agent ready your blo your website is. Um which is the it's agentready.com. Right. I I checked out this I put my own personal website into the readiness um checker and I got 20. Not very good. But what I love about um this this tool here is it tells you exactly how to improve your score. I just included here salels.io. Selo it's is the engineering manager of the team that actually did this. So I put his blog and uh his blog is actually pretty great. Uh yeah, people can definitely play around and I really like this feature where you can improve your score and just copy all the instruction and put in your your uh agent of of choice and it will improve your website giving the feedback uh which is great. Yeah, very meta. Very meta. Exactly. Uh anything we should mention about the Friday as well? One story I I I I I I personally was um really fascinated by is um uh shared dictionaries uh compression that keeps with the agentic web. Um, so this one I I think it it actually was uh uh while we were still working like a week before agents week, that's when like this announcement came in because it was like a relatively new concept. I think like um recently was enabled within like the Chrome browsers, but uh I could be wrong about that. What what I think is really cool about um this particular announcement is that um you know especially as more agents are writing more of our web applications and um you know applications web apps in general are just getting bigger and bigger um we need a way to kind of like uh uh deliver these these applications to the customer as efficiently as possible. And so with a comp um compression uh dictionary essentially we're comparing what kind of uh parts of the website that the client already has already has cached and we're only delivering the next time you come to a website we only need to kind of like give you the difference even if it's like the website has updated and so it g it makes your uh web apps be served like faster and also more like uh efficiently and so I think this one will really be impactful in terms of um you know the next era of the web where both agents are writing more uh code and also because agents will also be traversing the web and and and visiting websites as I don't resist. Let me share this tweet from Alec Rivik that actually wrote the blog about the difference in terms of uh uh this this uh feature this product can do like 99.5% smaller compression just got a memory uh so it's it's uh that's really important in this day and age like savings efficiency all of that will be more relevant now that agents are doing more in terms of uh what they can achieve right Uh anything we want to mention? Another actually that had some attention was the introducing agent memory uh managed service that gives AI agents persistent memory allowing them to recall what matters forget that what doesn't uh and get smarter. That's a that's also a perspect a good perspective right yeah I I I can speak a little bit about agent memory. Um so I think it's super interesting because there's just so much discussion um around how you know when you're building an agent how do you manage um its context right some people call it like context engineering and like there's been a lot of ways in which people are discussing like how memory should uh work within a harness and so um what we have here is like uh we we have kind of like a fully managed like out of the box product that you can directly leverage to give your agent everything it needs to manage its uh context with like an end user. So it's uh one component about it is like via retrieval and search. So like if you um uh you know have uh conversations you previously talked about before then it it can kind of like give you that uh search through that previous context and fetch what it needs to use it within you know a given prompt or task. Um, and then it also has like built-in like uh compaction or some people call it like autodreaming which helps you like um uh as you're having more conversations with the user like the previous context is summarized um so that it can be leveraged for f in the future without kind of cramming up the context window. I think uh agent memory will be really useful for anyone trying to build any sort of agents. These are the announcements so far. anything we could mention about the Monday ones? The ones coming on Monday, I I think some of them will be get doing a little bit more of a deeper dive around how Cloudflare sells how we use how we how we've set up our own like AI and like aentic coding stack. I think that you know so many c companies out there are trying to do this themselves. Everyone's trying to deploy agents. Everyone's trying to figure out how do we make our engineering team or our entire team generally more productive with agents. It takes a lot of you know I think everyone's kind of trying to figure this out like once and you know what are the best practices and I think you know we're just gonna we'll be sharing some of like what we do the things that we've learned from trying to set this up what are what are the best practices that we found. So I think you know if you're trying to do something like this within your own organization that'll be really uh really interesting and and good to read. Absolutely. And to be honest, the way we do things is there's a lot of things that agents bring that people don't control and don't understand too much. Security elements in place. May that be we talked about container sandbox, but even AI gateway, things like that. All of those things as an ecosystem will bring more protection, will bring actually at ease if if you know that you can explore things in a safe environment and play around with it knowing that you're safe there if you do those type of things. and those examples, sharing those examples is always interesting for sure. So, let's stay tuned for that. When this is out, those vlogs will be already be out. So, stay tuned for that for sure. Um, any in terms of the feedback, any surprises that you got from the week? We already mentioned some of the popular ones, but anything that surprised you in terms of even customers being really eager for some of the announcements? I really enjoyed watching like seeing everyone on on Twitter and stuff like try out the is my site agent ready and posting their own scores and I think you know people were posting some like pretty low scores but then also some people who had like a very you know like 100% agent readiness. Um and I think that just like speaks to like you know the the kind of wide range that that like how ready people are for the for this new era. But yeah, I always get a like anything I think like I always get a kick of these things that are like interactive and you could kind of like try it out on your own site. Yeah, I I I this one isn't like I guess related to a specific announcement, but I saw a couple threads on Twitter where people are like, "What is going on with Cloudflare this week?" Like some people might not have been fully clued in. We have like agents innovation week and so they're just seeing like a bunch of announcements come out and and they're like, "What's what's in the waters?" You know, what what are they feeding the people at Cloudflare? So, uh I thought that was kind of funny as well. for for sure that score I I was really close because the team is here in Lisbon Andre Jesus actually I usually work next to him and I saw even on Twitter today people sharing their bad results and like 20 minutes after sharing hey I just gave this to my agent and now the score is like great so how fast you can actually do improvements these days it's quite astonishing for sure and this is a good example observability in this case also it's it's helpful for what you want to achieve specifically. So definitely a a great one. I don't resist I won't resist but there's call for radar also has some insights related to that announcement adoption of AI agent standards uh for example how many are using robot.ext or sitemap AI rules in robot.ext text. There's a bunch of cool trends here that people can monitor on radar. Uh that is pretty cool to be honest. And URL scanner actually has also a a feature related to uh that those types of scans which are great that people can explore as well. Um oh well uh this has been great. Uh anything we want to add for folks to be attentive for what's coming? I guess still stay tuned for Monday and Tuesday because we still have a few things up in the pipeline and even then like beyond I think we're we're we're always cranking out new features, new products and so even though this is the official end of like agent week, I don't think we'll ever really stop continuing building and hopefully giving everyone what they need to for their future like agent workloads. For sure. Ming, any thing you want to add? Um, yeah, you know, I think this is this is agents week, but I don't know, 2026 is probably like agents year, uh, to be honest. And yeah, a lot of kind of like what, uh, at least kind of I know for my team and I'm sure for a lot of other teams on Cloudflare, what we're thinking about for the rest of the the whole year is like how do we build better tools for people building agents and uh, so yeah, I expect a lot more like agent related things coming out the the rest of the year as well. For sure. To be honest, it's April and it seems like already a full year of announcements of things and we're just like not in the middle of the year still. So, what a year. That would be a a good expression there. Thank you for doing it and uh stay tuned. Geek out. Thank you. And that's a wrap. So, before we go, here's a video from my colleague Zeke from our developer team. So, Zeke made this deep fake video. It's also a deep fake audio because it's mimicking Zeke's voice, but it's not Zeke's voice specifically. Zeke use many models for different uh things, video, audio generation, dialogue, sound effects, a bunch of things. It uses replica models available on Replicate. Replicate is a company that call acquired in late 2025. So, it's definitely worth a watch. It only covers four days of agents week uh up until Thursday, not Friday. But it's actually very cool summit. not only in terms of video, but also in terms of what he says. So, uh, here it is. Here's Zeke. It's agents week at Cloudflare. The week's not over yet, but there's already too much to keep up with. Zeke couldn't keep up, so he deployed me as his deep fake to give you the scoop. Let's cover the highlights. Here's the thing about agents. Every traditional app serves many users from one server. Agents flip that. One user, one agent, one task. scale that to millions of people and you need millions of simultaneous sessions. Containers can't do that. Not at a price anyone would pay. So, Cloudflare built something different. Project Think is the next version of the agents SDK. Your agent can crash and recover. It can hibernate when idle and cost you nothing. It can spawn sub agents and escalate from a lightweight isolate to a full container when it needs one. 10,000 agents on containers means 10,000 always on instances. On durable objects, maybe a hundred are active at any moment. That's the difference. Sandboxes are now generally available to everyone. Your agent gets a real computer terminal code interpreter live preview URLs, secure credential injection. Figma is already using them for Figma make. Durable objects got facets. That means every AI generated app can have its own SQLite database supervised by your code. If you're building a platform where users can vibe code their own apps, each app gets its own isolated state. Browser Run lets your agent control a headless browser with a live view. If the agent gets stuck on a login page, it hands off to a human. The human logs in, the agent picks back up. You get session recordings, direct CDP access, and support for up to 120 concurrent browser sessions, voice agents, got a new SDK, wrap any agent class with with voice, and it can hear and speak in real time over websocket. Built-in speechtoext and texttospech providers, React hooks included. The Cloudflare email service entered public beta. Agents can send and receive email natively from workers. There's an on email hook in the agents SDK so your agent can receive a message, do hours of background work, and reply when it's done. AI search shipped as a retrieval primitive. Create search instances on the fly. Upload documents and query with hybrid semantic and keyword search. One binding built-in storage and vector index. No external services needed. The registar API is now in beta. Agents can search for available domains, check pricing, and register them programmatically. Three API calls, a few seconds. It's already wired into the Cloudflare MCP server. Artifacts is a versioned file system that speaks Git. Create a repo per agent session, fork sessions, time travel through state, import from GitHub, clone with standard Git tools. Uh, it's built on durable objects and it's heading to public beta next month. The AI platform unified 70 plus models from 12 providers behind one API and one bill. The replicate team is now fully merged into Cloudflare. You can bring your own model via Cog and AI gateway now buffers streaming responses so if your agent crashes midstream, it can reconnect without repaying for the inference. Cloudflare built a custom inference engine in Rust and shipped pre-filled decode disagregation, speculative decoding, and crossGPU KV cache sharing. The result, large language models that are fast enough for real-time agent loops running on Cloudflare's own GPU fleet. There's a new unified CLI, run npx.cf, to manage any Cloudflare product from your terminal. It's backed by a new TypeScript schema system that generates CLI commands, config bindings, and docs from a single source. Plus, a local explorer that lets you inspect your local dev state for KV, R2, D1 durable objects and workflows. Managed oath lets you flip a switch and make any internal app behind Cloudflare access agent ready no code changes. And Cloudflare Mesh wires up private networking between your devices, servers, agents, and workers so everything can talk to each other securely without a VPN. That's 20 blog posts and it's only Thursday. One more day to go. I'm Zeke's deep fake. You stay classy developers and good luck keeping