Defang V3 - One command deploys your app to any AWS or GCP account.
by•
Deploying should be as easy as building. And now it is!
Defang is the bridge between your coding agent and your cloud. And with V3, Defang is better than ever. One command deploys straight from your IDE to your AWS or GCP account. Spin up dev, staging, and prod directly from a Compose file. Effortlessly deploy your CrewAI, LangGraph, n8n, and more. Automatically catch errors and fix them. Integrate seamlessly into your CI/CD pipeline.
Your cloud. Your infrastructure.
Free for open source!


Replies
Defang
Hey everyone, Lio here - CTO and co-founder of Defang.
I always wished that deploying would be as easy as building. That's been our north star for three years. V3 is where it becomes real.
One command. Your choice of cloud. Production-ready infrastructure. The CLI catches errors and fixes them. Spin up dev, staging, and prod environments from a single compose file. Deploy straight from your editor, Cursor, VS Code, or Claude Desktop.
Built for the AI agent era: CrewAI, Mastra, LangGraph, n8n, or any other containerized app. One line to enable LLM, whether that’s Bedrock or Vertex AI.
Your customer wants your agent in their own cloud? One command. Your app, their data, in their infrastructure. Free for open source.
To everyone who supports us today: 2 months of Defang Pro free. We're also selecting 10 supporters for lifetime Pro access. And we’re free for public GitHub repos.
Thank you for supporting us.
@lio_lunesu congrats on the launch. Does this replace Vercel?
@lio_lunesu @austin_heaton Vercel is a PaaS, basically an AWS wrapper that makes deployment easy, but your code runs on their infrastructure. Defang makes it just as easy to deploy, but directly to your own AWS or GCP account. Same simplicity, but you keep full control and ownership.
Defang
@austin_heaton Like Gabriel mentioned, Vercel runs your code, and you'll be locked in to Vercel, unless you make significant changes to your architecture and code. With Defang we want you to be able to deploy to whichever cloud you choose (hence the name!). Of course, it'll take time to support more clouds, but today we support AWS and GCP, with limited support on DigitalOcean.
@lio_lunesu Congrats on the launch Lio. Whats been the biggest learning in building V3?
Defang
@zolani_matebese Hi Zolani. The biggest learning is that people love the Vercel/Render/Fly/Railway PaaS experience, but that comes with lock-in, making it hard to leave the platform. Unfortunately, the hyper scaler clouds could give you choice, but that comes with a worse developer experience (and of course they also have their own APIs and what to lock you in too). What folks are really looking for is for cloud providers to become a "utility", just something that's there and you can use, with the ease that's expected from dev tools in 2026.
Hey all! I'm part of the Defang team as well, and just wanted to leave a little story here too.
I remember meeting Lio in March 2022. Our mutual friend had put us in touch, saying that we were both working on a similar problem:
It's too damn hard to deploy to the cloud.
I was approaching it from a very high level: a framework that provided high level primitives that could address many use cases, and some underlying infra provisioning code to make sure those high level primitives played nice in cloud providers.
Lio was working from the other direction: I remember him talking about how to manage pretty low-level DNS stuff and making sure it was easy to deploy and extremely resilient.
We started grabbing coffees more frequently, talking about Kubernetes, AWS, and all kinds of other stuff we were both dealing with as we worked on these problems separately.
Man, did we have some wacky conversations about managing k8s, how to map out a local development story with a cloud deployment story, exploring stuff like making queues and serverless work nicely in a local and cloud environment, across use cases.
He was building what became Defang.
Eventually I met the rest of the team as they settled on Docker's "Compose" file as a standard to define an app that would work locally and in the cloud.
To me that sounded like magic: I started using Docker Compose back in 2016 and had probably searched for "scalable Docker Compose to Cloud" a few times per year, but to no avail (Swarm was always clunky and the CF templates always crashed on me).
Defang's approach was different, too: the idea wasn't to shove everything into an expensive Kubernetes cluster, but to use the cloud provider's native container runtimes, like AWS ECS or GCP Cloud Run to the extent possible, which has all kinds of benefits.
Eventually I helped build the first version of the Defang Portal and later on joined the team.
The "compose to cloud" flow has been working for a while, but it worked best locally, or if you were already familiar with your cloud provider's tooling.
This release changes that.
I've been testing our flows and I love this: you can log into Defang, AWS, and GitHub and use Defang to connect them all in the browser to launch an app that uses queues, databases, LLMs and more without ever leaving your browser.
And it's all running in your own AWS account.
You can also use our MCP server or our CLI agent to help you deploy your app from your machine to the cloud: you don't need to be an infra expert to deploy something pretty complex.
I'm really excited about this release: this is the product I've been looking for for almost 10 years, and I think a lot of you would actually love it.
I always looked for solutions to avoid deploying to AWS (read: used AWS wrappers lol). The first time I tried to deploy my app to AWS with Defang I fully thought it didn’t work because it was too straightforward. What a delight. I wouldn’t even begin to try with AWS any other way.
Really excited to try it out, looks like a lot more polish has gone into the product to make it even easier to get set up and keep shipping. Would highly recommend to all my friends.
Great work, team!
@wzich thanks for the kind words! I hope you like the new features!
I've known co-founder Lio for 15 years and he's always been the top maker I know. With defang he's surrounded himself with a top notch technical team that have built an extraordinary product with a laser focus on the developers needs to deploy production-quality infrastructure as an extension without an army of DevOps to create and maintain it. Defang ought to become the way we deploy all our apps.
@rngadam thanks for the kinds words Ricky! Grateful for your support :)
Hi, congrats on your launch!
What is not clear for me, though, how is that different from, say the AWS "eb deploy" command?
Thanks!
@seacat Hey there Kath!
A few things really:
Defang apps are defined using Docker Compose files. So you can essentially run `docker compose up` to run your app locally, and then `defang compose up` to run them in your cloud account, including provisioning databases, caches, ssl certificates, dns, etc.
You can deploy to AWS or GCP with the same command (and working on adding more cloud provider). For some people this is useful if they are deploying to their client's cloud accounts. For others it's useful to be able to switch clouds for pricing reasons (though you still have to handle data transfer manually)
Integration with AWS Bedrock and GCP Vertex. With a simple flag we'll make sure to provision the correct security configurations so that your services can easily connect to LLMs, embedding models, etc.
Deployment modes allow you to specify whether you want to deploy a low cost version of your app, a balanced version, or a high availability version. We use that signal across your apps services to decide whether to use spot instances, db failovers, nat gateways, and more.
There are loads of other things that I think we do that make deploying with Defang simpler, but I think those are some of the big ones! Hopefully that's helpful :)
@seacat You're one of our 10 selected supporters winning 1 year of Defang Pro free! 🎉
Email us at i-gabriel.lunesu@defang.io to claim your reward.
I’ve spent an unreasonable amount of time trying to avoid deploying directly to AWS—usually by hiding behind wrappers and abstractions. When I first deployed an app to AWS using Defang, I genuinely assumed something was wrong because it was too simple. That moment of confusion quickly turned into delight.
Defang makes something notoriously complex feel almost invisible, and that’s rare. V3 clearly shows a lot of thoughtful polish—getting set up is easier, and staying in the flow while shipping feels much more natural.
Honestly, I wouldn’t consider approaching AWS any other way now. Highly recommend this to anyone who wants the power of cloud infra without the usual friction. Fantastic work by the team.
@gabriel_lunesu @Defang
@gabriel_lunesu @anudhyan_datta Thanks for the kind words! We're glad we could get you deploying to AWS so easily 😃
Big congrats to the team! I love how you're simplifying the 'Compose to Cloud' flow. Regarding the managed LLM integration (Bedrock/Vertex): Does Defang automatically handle the IAM permissions for my agent to reach those models, or do I still need to configure the underlying roles manually in the GCP console?
@santanu_kumar_das yes! Adding the `x-defang-llm` flag will automatically configure IAM/Permissions for your service to access Bedrock or Vertex.
@santanu_kumar_das Congrats, you're one of our 10 selected supporters winning 1 year of Defang Pro free! 🎉 Email us at i-gabriel.lunesu@defang.io to claim your reward.
I used to go out of my way to avoid deploying directly to AWS — usually relying on various wrappers to make it less painful. The first time I deployed with Defang, I honestly assumed something had gone wrong because it felt too easy. Turns out, it really was that simple. Such a great surprise. At this point, I can’t imagine doing AWS deployments any other way.
ALTAR
Really great work and exactly what I've been looking for! Not coming from a dev/ops background it's always been at least a bit bewildering if not downright frustrating trying to deploy and scale new products - even more so if we need to manage or migrate between multiple different clouds. Defang saves so much time for me and the rest of our team so we can all focus more on building.
@nz_altar Thanks Nelson! That frustration is exactly why we built Defang. Deployment shouldn't require a DevOps background, and managing multiple clouds shouldn't feel like a second job. Glad we're helping your team stay focused on what matters: building.
You're one of our 10 selected supporters winning 1 year of Defang Pro free! 🎉 Email us at i-gabriel.lunesu@defang.io to claim your reward.